historicaldevelopment - harvard university · evidence,for how one important historical...

CHAPTER SEVEN

Historical Development

Nathan NunnDepartment of Economics, Harvard University, NBER, and BREAD, 1805 Cambridge Street, Room M25,Cambridge, MA 02138, USA

Abstract

This chapter surveys a growing body of evidence showing the impacts that historical events canhave on current economic development. Over the past two decades historical persistence has beendocumented in a wide variety of time periods and locations, and over remarkably long time horizons.Although progress continues to be made identifying and understanding underlying mechanisms, theexisting evidence suggests that cultural traits and formal institutions are both key in understandinghistorical persistence.

Keywords

Persistence, Colonialism, Institutions, Norms, Culture, Path dependence

JEL Classification Codes

H11, N00, O10, 050, P51, R58, Z13

7.1. INTRODUCTION

In recent years,a new dynamic,empirical literature has emerged examining whetherhistorical events are important determinants of current economic performance.1 Theorigins of this literature can be traced to three lines of research that began approximatelya decade and a half ago: Engerman and Sokoloff (1997, 2002), La Porta et al. (1997,1998), and Acemoglu et al. (2001, 2002). Although each line of research had differentmotivations,what was common to them was that each provided an analysis,and supportingevidence, for how one important historical event—Europe’s colonization of the globe—was important for long-term economic growth.

Since this time, the literature has developed in a number of ways. Most notably,other important events have also been examined. These range from systems of laborcoercion,Africa’s slave trades, medieval long-distance trade,Atlantic trade, the ProtestantReformation,overseas missionary work,the French Revolution,the Mexican Revolution,the forced opening of China’s treaty ports, the adoption of new food crops during the

1 See Nunn (2009) and Spolaore and Wacziarg (forthcoming) for recent reviews of this literature.

Handbook of Economic Growth, Volume 2A © 2014 Elsevier B.V.ISSN 1574-0684, http://dx.doi.org/10.1016/B978-0-444-53538-2.00007-1 All rights reserved. 347

http://dx.doi.org/10.1016/B978-0-444-53538-2.00007-1

348 Nathan Nunn

Columbian Exchange, the adoption of the plough, the invention of the printing press,the Neolithic Revolution, and various environmental catastrophes.

The typical study involves the collection and compilation of new and impressive data.Although this in and of itself is an important contribution, the real contribution is the useof the data to convincingly test hypotheses related to historical development. The mostenlightening papers are able to trace the full impacts of a historical event through time,while examining specific channels and mechanisms.

This chapter provides a summary of this new literature. As we will see, once onesurveys the progress made to date, it is impressive what the recent wave of quantitativehistorical studies has taught us about historical economic growth and development.

7.1.1 The Origins of the LiteratureThe origins of the historical development literature can be found in three sets of papers.What the three papers have in common is that they all examine European colonial rule.However, their motivations are very different.

The first study, written by economic historians Engerman and Sokoloff (1997), is ahistorical narrative, supported with descriptive statistics. In it they examine the impor-tance of factor endowments and colonial rule for the subsequent economic developmentof colonies within theAmericas.They argue that NewWorld societies that were endowedwith soil and climate suitable for growing lucrative, globally traded commodities, suchas sugar, tobacco, and cotton, developed plantation agriculture, and with it, the use ofslave labor. In the Spanish colonies, characterized by sizable indigenous populations andlarge reserves of gold and silver, forced labor was instituted.The use of slavery and forcedlabor resulted in economic and political inequality, both of which inhibited long-termeconomic development.

Interestingly, the other two seminal articles were not inherently interested in bet-ter understanding the history of European colonial rule. For example, the interest ofAcemoglu et al. (2001) was in testing whether domestic institutions are a fundamentaldeterminant of economic prosperity today.The interest of La Porta et al. (1997,1998) wasin identifying the causal impact of investor protection on financial development. Whatmotivated both studies to examine colonial rule is the fact that the historical episode pro-vides a source of variation in domestic institutions (in the case of Acemoglu et al.) and ininvestor protection (in the case of La Porta et al.). Both studies exploited European colo-nial rule as a natural experiment, focusing on a different dimension or characteristic thatwas argued to provide exogenous variation that they could use to identify their effect ofinterest.

La Porta et al. (1997, 1998) argue that because the legal tradition of the colonizerwas transplanted to the colonies, the identity of the colonizer had an important impacton the legal system that evolved and, in particular, on contemporary investor protection.In particular, they show that former colonies with a legal code based on Roman civil

Historical Development 349

law—these were the colonies of France, Spain, and Portugal—had weaker investor pro-tection and less financial development relative to former British colonies with a legalsystem based on common law.

Acemoglu et al. (2001) argue that a primary determinant of the form (and long-termimpacts) of colonial rule was the disease environment faced by potential European settlers.In temperate areas, like Canada,Australia, and the United States,European mortality rateswere moderate enough to facilitate European settlement on a large scale. In these areas,the Europeans brought with them their values and beliefs and developed European-likeinstitutions that emphasized the protection of property rights. In areas such as sub-SaharanAfrica, where European mortality was high due to diseases like malaria and yellow fever,Europeans did not settle. Instead, they engaged in an extractive strategy. Rather thansettling in a location, they set out to extract natural resources without regard for theconsequences. Arguably, this strategy was facilitated by a lack of property rights and othersimilar institutions. Motivated by this historical narrative, Acemoglu et al. used a measureof early settler mortality as an instrument for contemporary domestic institutions toestimate the causal impact of institutions on long-term economic development.

The analysis of the three lines of research showcased how insights can be gained byexamining economic growth and development from a historical perspective. Specifically,they showed how historical episodes can provide econometrically useful sources of exoge-nous variation. More importantly, they also showed that history matters and that it canhave long-term persistent impacts that continue to influence growth and developmenttoday.2

Following these early studies, a large number of subsequent papers have emergedexamining economic growth and development from a historical perspective. In the fol-lowing section, I begin an overview of this literature by first describing a number ofstudies that examine other dimensions and aspects of European colonial rule, the histor-ical event that has received the most attention in the literature. In Section 7.3, I thenturn to an examination of studies that have investigated the long-term impacts of otherhistorical events. These include the Columbian Exchange; various episodes of increasedtrade and globalization; episodes of warfare and armed conflict; expulsions and forcedpopulation movements; religious reformations; and important technological innovations.Following this, in Section 7.4, I turn to an important insight that has emerged from theliterature: geography has important impacts on development today working through itsimpacts on historical events.

After having surveyed the evidence for the importance of history for contemporaryeconomic development, I then turn to causal mechanisms. In Section 7.5, I summarize

2 As with any seminal paper, extensions, comments, criticisms, criticisms of comments, criticisms of crit-icisms, etc. soon emerged. In an effort not to get lost in the weeds, I do not discuss these papers here.See for example, Easterly and Levine (2003), Glaeser (2004), Olsson (2004), Rodrik et al. (2004),Austin(2008),Albouy (2012), and Acemoglu et al. (2012).

350 Nathan Nunn

the evidence that has been uncovered for the relative importance of various channelsof persistence, including multiple equilibria and path dependence; domestic institutions;cultural values and beliefs; and genetic traits.

The final two sections of the chapter,Sections 7.6 and 7.7,discuss unresolved questionsin the literature and offer concluding thoughts.

7.2. EUROPEAN COLONIAL RULE

7.2.1 AmericasThe studies that examine the impacts of colonial rule in the Americas tend to focuson testing the hypothesis that initial endowments affected the extent of economic andpolitical inequality,both of which were detrimental for long-term economic development(Engerman and Sokoloff, 1997). In a followup study, Engerman and Sokoloff (2005)provide additional evidence for their hypothesis by documenting a positive relationshipbetween economic inequality and political inequality, measured by the inclusiveness ofvoting rights. Sokoloff and Zolt (2007) document a link between inequality and lowertaxes on wealth and income and less spending on public goods such as education.

While the evidence put forth by Engerman and Sokoloff in support of their hypothesisprimarily takes the form of historical narrative and descriptive statistics, a number ofstudies have undertaken more formal tests of their hypothesis. Bruhn and Gallego (2012)examine variation across 345 regions within 17 countries from North and SouthAmerica.They identify a strong negative correlation between long-run development and initialcolonial specialization, in what they call “bad” activities, which Engerman and Sokoloff(1997) argue display economies of scale and therefore relied heavily on exploited labor,e.g.sugar, coffee, rice, cotton, and mining. They provide additional evidence, also consistentwith Engerman and Sokoloff (1997), that other activities, like subsistence farming, cattleraising, or manufacturing, are not negatively related to long-term development, unlessthere were large native populations that could potentially be exploited in the productionprocess.

Naritomi et al. (2012) provide evidence consistent with Bruhn and Gallego (2012),but focus on Brazil and two commodities, gold and sugar.They examine variation acrossapproximately 5000 Brazilian municipalities and quantify each municipality’s historicalinvolvement in the gold boom (during the 1700s) and the sugarcane boom (1530–1760).The authors show that the municipalities that experienced the sugar boom have greaterland inequality today, while municipalities that experienced the gold boom have worsedomestic institutions today.

The key mechanism in Engerman and Sokoloff ’s hypothesis is inequality, both eco-nomic and political. A number of studies provide evidence that calls into question theirassertion that greater economic inequality is associated with greater political inequalityand less development. Dell (2010) examines the mita forced labor system, which wasinstituted by the Spanish in Peru and Bolivia between 1573 and 1812. The mita system


required that over 200 communities supply one seventh of their adult male populationto work in the silver mines of Potosí and mercury mines of Huancavelica. The studycombines contemporary household survey data, geographic data, and data from historicalrecords; and uses a regression discontinuity estimation strategy to estimate the long-termimpacts of the mita system. Dell’s study exploits the fact that the boundary of the mitaconscription area was clearly defined and that other relevant factors likely vary smoothlyclose to the mita boundary. Because of this,comparing the outcomes of mita and non-mitadistricts very close to the border provides an unbiased estimate of the long-term effectsof the mita. The study finds that the mita system had an adverse effect on long-termeconomic development. All else equal, former mita districts now have an average levelof household consumption that is approximately 25% lower than households in formernon-mita districts. The study finds that a significant proportion of the difference can beexplained by lower levels of education and less developed road networks.

Dell argues that the underdevelopment of mita districts was due to an absence oflarge haciendas. These haciendas lobbied the crown for public goods, like educationand roads, and provided these goods directly. Therefore, in contrast to the Engerman-Sokoloff hypothesis, she finds better long-run development outcomes in locations withlarge haciendas and greater inequality (not less).

Acemoglu et al. (2008) also question Engerman-Sokoloff ’s inequality hypothesis.Theauthors first examine municipalities within Cundinamarca, Colombia and show that late19th century land inequality is positively associated with late 20th century secondaryschool enrollment. They further question the presumption that economic and politicalinequality go hand in hand. After constructing a measure of political inequality usingdata on the identity of mayors for 4763 appointments held by 2300 different individualsbetween 1875 and 1895, they show that economic inequality and political inequality arenot positively correlated. In fact, they argue that greater land inequality was better forlong-term development because the landowners provided greater checks on the actionsof the political elite.

Examining variation across US states and counties and across countries within theAmericas,Nunn (2008b) also considers the role of inequality. Although he does find that,consistent with Engerman and Sokoloff, there is a negative relationship between slave useand current income, he fails to find evidence that inequality is the intervening channel.Although past slave use is positively correlated with historical and current inequality,controlling for historical land inequality does not reduce the negative impact of slaveryon current income. Further, there is no relationship, either in the past or today, betweeninequality and income.

7.2.2 AsiaEarly European contact with India occurred through overseas trade, beginning in 1613.Colonization of the subcontinent occurred through a number of battles. Beginning withthe Battle of Plassey in 1757, the British East India Company (EIC) gained control

352 Nathan Nunn

Figure 7.1 Directly ruled British districts and princely states within the Indian Empire. Source: ImperialGazeteer Atlas of India, Plate 20.

of Bengal and Bihar, and by the early 19th century, the British controlled large partsof the Indian subcontinent, the other portions being the “princely states.”The BritishEIC continued to annex the princely states during the 19th century until the mutiny ofthe British EIC’s army (the Sepoy Mutiny) in 1857. After this, the British governmentruled the subcontinent, establishing the British Raj.

The long-term impacts of British control on the Indian subcontinent have beenexamined empirically in a series of recent papers. Iyer (2010) examines the long-termimpacts of direct British rule versus indirect rule—i.e. the princely states. The portionsof the subcontinent under the two forms of rule (in 1909) are shown in Figure 7.1.

Looking across 415 districts, Iyer (2010) estimates the effect of direct British ruleversus indirect British rule on investment in agriculture and agricultural productivitytoday. To help uncover causal estimates, she exploits the Doctrine of Lapse, a Britishpolicy that was in place between 1848 and 1856, that stated that a native ruler’s adoptedheirs were not to be recognized by the British government. This allowed the British to


annex several states where the native ruler died without a natural heir. Iyer instrumentsfor direct British rule using a district-specific indicator variable that equals one if theruler died without a natural heir between 1848 and 1856, the period when the Doctrineof Lapse was in place. Examining the subset of states that had not yet been annexedby 1848, she shows that the IV procedure estimates no statistically significant differencebetween directly ruled districts and the princely states. This is in contrast to the OLSestimates which suggest that direct British rule is positively associated with agriculturalinvestment and productivity. The most likely explanation for the difference is that theBritish annexed the most productive parts of the continent, which also had the greatestgrowth prospects.

In subsequent analysis, Iyer (2010) examines other contemporary outcomes, includingthe availability of public goods such as health, education, and roads. She continues to findthat the IV estimates of the impact of direct British rule on public goods provision arelower than the OLS estimates, again suggesting that the British selected the “better”states, with greater long-run growth potential. In addition, for many of the public goodsoutcome variables, the IV estimates suggest that British rule actually exerted a negativelong-term effect, a finding that is consistent with earlier research by Banerjee et al. (2005)who examine an even larger set of 27 different public good measures.

Other research, rather than examining differences between directly and indirectlyruled districts, looks at variation within the directly ruled districts of India. Banerjee andIyer (2005) show that differences in the institutions initially implemented by the Britishhad long-term growth effects. In particular, they examine the different revenue collectionsystems that were established and compare districts where revenue was collected directlyby British officials to those where revenue was collected by native landlords. They findthat after independence, districts with non-landlord systems have higher levels of health,education, and agricultural technology investments relative to landlord systems.

To determine the extent to which this correlation is causal, the authors exploit thefact that in the parts of India conquered between 1820 and 1856, non-landlord revenuecollection was implemented. They argue that the historical determinants of the form ofrevenue collection are orthogonal to district characteristics and determined primarily bythe date of British conquest, which they use as an instrument for the revenue collectionsystem. Their IV estimates are consistent with the OLS estimates.

Overall, the existing body of evidence for India suggests that British control of thesubcontinent, through the extent to which colonial policies took the form of direct rulevs. indirect rule, had lasting impacts on long-term economic growth.

7.2.3 Island ColoniesIn a novel study, Feyrer and Sacerdote (2009) examine the experience of Europeancolonial islands of the Pacific, Indian, and Atlantic Oceans. Although one could argue thisis a somewhat obscure set of colonies to examine, the question they attempt to answer is

354 Nathan Nunn

of general interest. Their motive for looking at their particular sample stems from theirinterest in the obvious, but difficult-to-answer question of whether colonial rule, onaverage, was good or bad. In particular, is a longer period of colonial rule associated withbetter or worse long-term economic growth?

Feyrer and Sacerdote (2009) argue that, for islands, the date of discovery was deter-mined in part by its location relative to prevailing wind patterns and that these windpatterns most likely do not affect long-term development through channels other thanthe island’s date of discovery. They argue that the wind vectors surrounding an islandcan be used as instruments to estimate the causal effect of the length of colonial rule onsubsequent development. Their baseline set of instruments, which are constructed fromsatellite imagery data, reported monthly on a one-degree by one-degree global grid,include the annual mean and variance of monthly east-west wind vectors.

Their first-stage estimates show that stronger westerly winds are associated with earlierdiscovery and more years under colonial rule. According to their 2SLS estimates, thelength of colonial rule has a positive effect on per capita income in 2000. In other words,conditional on being a colony (within their sample), a longer period of colonial rule wasbetter for economic development.They are quick to point out,however, that their resultscannot address the question of whether or not the island colonies are better off becausethey were colonized.

7.2.4 AfricaA number of studies that examine the impacts of colonial rule withinAfrica find evidenceof long-term impacts that persist until today.This is perhaps surprising since, for the vastmajority of the continent, the period of colonial rule was short relative to the rest ofthe world. Africa was the last continent to be colonized, with the Berlin Conference of1884/1885 marking the beginning of large-scale colonial rule within Africa.

An important source of evidence documenting the long-term impacts of colonial rulewithin Africa is Huillery (2009). In the study, her analysis combines data from historicaldocuments from archives in Paris and Dakar with household surveys from the 1990s. Sheshows that looking across districts in French West Africa, there is a positive relationshipbetween early colonial investments in education, health, and infrastructure and currentlevels of schooling, health outcomes, and access to electricity, water, and fuel. Most inter-estingly, she provides evidence of persistence that is specific to a particular public goodand outcome. In other words, she finds that greater education spending during the colo-nial period is associated with more education in the post-colonial period, but not betterhealth outcomes or more infrastructure. Similarly, she finds more infrastructure invest-ment during the colonial period is associated with greater access to infrastructure today,but not with the other outcomes;and more health investments during the colonial periodare associated with better health outcomes today, but not with the other outcomes.


While the exact mechanisms behind this somewhat extreme form of persistence arenot yet well understood, she does provide evidence that early investments subsequentlylead to more of the same investments.An alternative explanation is that persistent omittedfactors,which have impacts that are public-good specific,are driving the results. However,her analysis undertakes a number of strategies to rule out this explanation, includingmatching districts based on geographic proximity.

A commonly cited adverse consequence of European colonial rule in Africa is thatit resulted in the creation of country boundaries that paid little or no attention to pre-existing kingdoms, states, or ethnic groups. During the Berlin Conference of 1884/1885,the European powers divided among themselves lands that they had yet to explore, andlakes, rivers, and mountains that they had yet to discover. Although it has long beenhypothesized that one of the legacies of colonial rule in Africa may be the artificialnature of national boundaries that it created, it was not until recently that this assertionwas tested formally. Michalopoulos and Pappaioannou (2011) combine information onthe pre-colonial locations of 834 ethnic groups from Murdock (1959) with the currentboundaries of nation states and test for differences between ethnic groups that werepartitioned by a country’s border and those that were not. The authors create, for eachethnicity, two indicator variables. The first equals “1” if the ethnicity was partitioned bya border and greater than 10% of the area of the group lies on both sides of the border.The second equals “1” if the ethnicity was partitioned by a border but less than 10%of the area of the group is located on one side of the border. The examination of thetwo partition measures is motivated by potential measurement error due to imprecisionin the location of ethnic group as mapped by Murdock (1959). Even if borders do notsplit ethnic groups, measurement error will generate splitting. This form of partitioningis more likely to occur in the second (less than 10%) measure.

The authors examine two sets of outcomes at the ethnicity level: economic devel-opment, measured by the density of night-time lights, and the number of civil conflictsbetween 1970 and 2005. They find that partitioned ethnic groups are associated withlower economic development (as measured by lower light density) and more civil war.Both partition measures are statistically significant, although the magnitude of the lessthan 10% measure is often smaller, which is consistent with this variable being measuredwith greater error.Therefore, their findings confirm the conventional belief that colonialrule,because of the way it artificially divided up the continent between European powers,had detrimental impacts.

Another potentially adverse consequence of colonial rule—particularly the policy ofindirect rule—was that it often resulted in heightened hostilities between ethnic groups.In Rwanda,colonial policies intentionally deepened racial differences between the Hutusand Tutsis. The Census of 1933–1934 institutionalized the Hutu distinction, creatingidentity cards that reported individual ethnicity. In addition, an educational system with

356 Nathan Nunn

separate streams for Hutus andTutsis was implemented. Prior to the arrival of the Euro-peans, the distinction between Hutus and Tutsis was more one of class than of race, andwith much movement between the two groups. This is illustrated by the fact that it wasactually quite difficult for the Belgians to group individuals neatly into one of the twogroups. As a result, they developed the “Ten Cow Rule”. When there was doubt, if anindividual had more than ten cows, they were designated “Tutsi”; otherwise, they were“Hutu.”

Many scholars,and perhaps most notably Mamdani (2001),have argued that the ethnichostilities between the Hutus andTutsis,culminating in the 1994 genocide,have their rootsin Belgian colonial rule. However, there is far from full consensus on this issue.Vansina(2004), relying primarily on oral evidence and early written accounts, argues against thisview. He instead argues that Hutu and Tutsi identities arose in the 17th century andbecame further entrenched during the 19th century, both of which occurred before thecolonial period.

7.3. OTHER IMPORTANT HISTORICAL EVENTS

The literature’s initial focus on European colonial rule was perfectly natural giventhat the event was one of the most important in human history and arguably the singlemost important for shaping the current distribution of the world’s income. However,more recently, researchers have begun to turn to other important historical events toempirically examine their long-term impacts and importance for economic developmenttoday. I now turn to a discussion of these studies.

7.3.1 The Columbian ExchangeThe Columbian Exchange refers to the transfer of crops,disease,ideas,and people betweenthe Americas and the rest of the world following Christopher Columbus’s voyage to theNewWorld in 1492.The exchange brought diseases that decimated the Native Americanpopulations. It introduced the Eastern Hemisphere to a variety of new plants that werewidely adopted, including tomatoes, the white potato, the sweet potato, cassava, corn,chillis, peppers, cacao,vanilla, and tobacco. In addition,Europeans were introduced to thechincona tree, which produces quinine, a prophylactic against malaria. The New Worldalso provided abundant fertile land that could grow valuable Old World commodities,such as sugar and cotton.3

A number of papers have documented the impacts of the transfer of new foods fromthe Americas to the rest of the world. Nunn and Qian (2011) estimate the impact of theintroduction of the potato to Europe, Asia, and Africa. Since the potato was calorically

3 See Grennes (2007),Nunn and Qian (2010), and Mann (2011) for further descriptions of the ColumbianExchange.


and nutritionally superior to Old World staples like wheat, barley, rye, and rice, for theparts of the Old World that were able to adopt the potato, its diffusion from the NewWorld resulted in a large positive shock to agricultural productivity. Using a difference-in-differences estimation strategy, the authors compare differences in population growth,urbanization, and adult heights before introduction relative to the population growthafter introduction between locations that were able to adopt potatoes and those that werenot. A location’s ability to cultivate potatoes is measured using GIS-based climate andsoil data from the FAO.

At the country level, they find that the introduction of the potato had a positiveimpact on total population and urbanization rates. Consistent with their urbanizationfinding, they also find that city growth, both globally and within Europe, was positivelyimpacted by the potato. In an attempt to examine mechanisms more closely, the authorsexamine height data from France. They show that after the diffusion of the potato toFrance, individuals born in villages that could cultivate potatoes were between one-halfand three quarters of an inch taller as adults.4

Other studies have also examined the impacts of the post-1492 diffusion of other foodcrops from the Americas. Chen and Kung (2012) examine the introduction of maize toChina and find that although maize had a large positive impact on population, there isno evidence that it spurred urbanization rates. Jia (forthcoming) examines the diffusionof the sweet potato in China. One characteristic of the sweet potato is that it is muchmore drought resistant than the pre-existing staple crops in China, rice and wheat. Heranalysis shows that prior to the sweet potato, there is a close relationship between theoccurrence of drought and peasant uprisings. After the diffusion of the drought resistantsweet potato, this relationship weakened significantly.5

The Columbian Exchange not only brought NewWorld crops to the OldWorld, butit also brought OldWorld crops to the NewWorld. For many crops the soils and climatesin the Americas were much more suitable for cultivation than in the Old World. Theprime example is sugar. Hersch and Voth (2009) calculate the welfare gains in Europefrom the increased supply of sugar that resulted from its large-scale cultivation in theAmericas. They also consider the welfare gains from the introduction of tobacco, thehighly popular New World crop, to Europe. They calculate that by 1850, the increasedavailability of sugar and tobacco had increased English welfare by approximately 6%and 4%, respectively. These results have important implications for our understanding ofEuropean well-being over time. It is generally presumed, based on real wage figures thatdo not account for new goods, that English welfare did not begin to improve until after

4 Cook (2013b) provides evidence of a complementarity between milk and potatoes. He finds that thepopulation effects of Nunn and Qian (2011) are greater among populations that exhibit lactase persistence(i.e. are lactose tolerant).

5 This finding is particularly interesting because it shows that the impacts of weather shocks differ dependingon the historical environment. This is a point that I return to below.

358 Nathan Nunn

1800 (Clark, 2005). However, according to Hersch and Voth (2009) welfare increasedsignificantly before this time from the availability of “new” goods that were a result ofthe Columbian Exchange.

7.3.2 International Trade and GlobalizationA number of studies have traced the impacts of increased international trade in variousperiods of time.An important insight that has emerged from this literature is that interna-tional trade, by altering the evolution of domestic institutions, can have important effectsfor long-term growth.6

The seminal papers making this point are Greif (1993 and 1994),where it is shown howthe development paths of two long-distance trading groups, the Maghribi and Genoese,have their origins in the manner in which trading contracts were enforced. Amongthe Maghribi, merchants relied on a collective enforcement strategy, where all merchantscollectively punished any agent who had cheated;while among the Genoese,enforcementwas achieved through an individual punishment strategy.These contracting institutions ledto the development of different institutions outside of the trading environment.While theGenoese developed a formal legal system and formal organizations to facilitate exchange;the Maghribis continued to rely on collective informal enforcement mechanisms likegroup punishment.

A number of papers have empirically examined the long-term impacts of increasedinternational trade and domestic institutions and long-term economic growth.The exist-ing evidence seems to suggest that trade can have very different impacts depending oninitial conditions and the specifics of the environment. The heterogeneous impacts ofinternational trade are clearly illustrated in the study by Puga and Trefler (2012), whichexamines medieval trade inVenice between 800 and 1350.They show that trade initiallychanged the balance of power, which enabled new merchants to push for greater polit-ical openness (e.g. the end of the hereditary Dogeship and the begining of theVenetianparliament) and better contracting institutions (e.g. the colleganza). These institutionalimprovements led to economic growth. However,over time,wealth was increasingly con-centrated in the hands of a relatively small number of merchants and this power was usedto block further institutional reforms and limit political access. Therefore, internationaltrade first led to the rise and then decline of growth-promoting inclusive institutions andeconomic growth.

The heterogeneous impacts of international trade can also be seen in the cross-section.For some parts of the world,evidence suggests that increased trade had beneficial impacts.Jha (2008) considers medieval India and shows that looking across cities within India,participation in overseas trade is associated with less religious conflict during the late19th and early 20th centuries. Jha addresses the endogeneity of the selection of medieval

6 On this point, see the recent survey by Nunn and Trefler (forthcoming).


ports by using the existence of natural harbors as an instrument for whether a coastal citywas a trading port and by using propensity score matching techniques.

According to his estimates, being a town that was a medieval trading port made it lesslikely that the town later experienced Hindu-Muslim riots. Using historical evidence,Jha argues that because Muslims provided access to the markets of the Middle East, thereturns to Hindu-Muslim cooperation were much higher in the towns connected to thisoverseas trade.As a result, institutions that supported exchange and a peaceful coexistencebetween Hindus and Muslims were developed.

The existing evidence suggests that international trade was also beneficial withinEurope. Acemoglu et al. (2005) examine the impacts of the Atlantic three-corner tradeon European institutions. They first show that the rise of Western Europe after the16th century was driven by the economic growth of countries (and port cities) heavilyinvolved inAtlantic trade.They argue that the primary benefit of the trade was not a directbenefit of the profits from the trade but an indirect impact that worked through domesticinstitutions. Profits from the trade shifted political power toward commercial interests andresulted in the development of growth-promoting institutions.They provide support fortheir hypothesis by showing that the Atlantic trade also resulted in better institutions,measured using an index of a country’s “constraints on its executive.” They also showempirically that among countries with access to the Atlantic trade,only for those that hadnon-absolutist institutions initially (i.e. in the 15th and 16th centuries) did trade generateimproved institutions. If the monarchy was too strong, initially, it simply monopolizedthe trade—as in Spain and Portugal—and this limited the benefits to the commercialclass and therefore limited institutional change.7

Beneficial impacts of international trade have also been estimated in 19th centuryChina. Jia (forthcoming) examines the impacts of increased trade due to the forcedopening to trade first imposed by Britain and later by the United States. The Treaty ofNanking (1842), which followed the Qing Dynasty’s defeat by Britain during the FirstOpiumWar,named five cities as treaty ports:Guangzhou,Shanghai, Fuzhou,Ningbo andXiamen (Fairbank, 1953). By 1896, 16 more treaty ports were added to the original five,with 28 more added between 1896 and 1911 (Tai, 1918;Wang, 1998).

Jia (forthcoming) examines a sample of 57 prefectures all with geographic access tooverseas trade over 11 periods between 1776 and 2000. She estimates a difference-in-

7 Another interesting source of heterogeneity related to the Atlantic trade is shown by Dittmar (2011). Heexamines the growth of port cities in Europe and shows that port cities that adopted the printing press bythe late 15th century grew significantly faster than those that did not.According to his estimates,once oneaccounts for the importance of the printing press, Atlantic port cities no longer appear to grow faster thannon-Atlantic port cities.These findings are consistent with Dittmar’s hypothesis that the printing press,bymaking print media more widely available, fostered numeracy, literacy, and innovations in bookkeepingand accounting,all of which were particularly valuable in cities with significant commercial opportunitiesdue to overseas trade.

360 Nathan Nunn

difference specification, controlling for prefecture fixed effects and time period fixedeffects. She finds that on average, prefectures with treaty ports experienced faster popula-tion growth during this period.The population increase occured prior to the Communistrevolution of 1949, as well as after China’s more recent period of increased openness after1980. Between 1958 and 1980,when China’s economy was heavily regulated, treaty portsdid not experience faster growth.

The clear exception to the beneficial impacts of trade and institutional and economicdevelopment lies in the experiences of Africa and (arguably) Latin America, followingthe Age of Discovery. As we have seen, in the Americas, specialization of production incommodities that Bruhn and Gallego (2012) classify as“bad”or“ugly”(recall Section 7.2)led to long-term underdevelopment. Further, as we discuss below, the impacts of theAtlantic trade within Africa were extremely detrimental. Within Africa, participation intrade meant warfare, theft, and banditry to supply slaves for export to the Americas. Asshown by Nunn (2008a) and Nunn and Wantchekon (2011), participation in the slavetrade had long-term adverse consequences.The parts of Africa that were heavily involvedin the trade today are poorer, have worse domestic institutions, and exhibit lower levelsof trust. We discuss these impacts in further detail below.

7.3.3 Warfare and ConflictHistory is filled with episodes of warfare and conflict. A number of recent studies showthat many instances of violence have had important effects on human history. The mostwell-known hypothesis regarding warfare and long-term development is Tilly’s (1990)hypothesis that an important determinant of the rise of Europe was interstate warfarewhich promoted the development of strong states. According to Tilly, beginning in theearly-modern period, warfare and interstate competition resulted in the developmentof centralized governments and institutions that were able to raise sufficient capital andmaintain large populations that could be used to wage war. In other words, according toTilly, war made states.

WhileTilly’s argument has been very influential,very few studies have actually formallytested any version of theTilly hypothesis. Aghion et al. (2012) provide evidence,which isvery much in the spirit of Tilly, that in the modern period an increased threat of warfare isassociated with increased education.The authors examine a yearly panel of 137 countriesfrom 1830 to 2001. Controlling for country fixed effects and year fixed effects, theyshow that education, measured by primary school enrollment and education reforms, ispositively associated with conflict in the previous 10 years and with a contemporaneousmeasure of the existence of a military rivalry with another country. In other words, theyfind evidence that war made education.

Proponents of the Tilly hypothesis have argued that it may also apply outside ofEurope. For example, Bates (forthcoming) and Reid (2013) argue that interstate conflictbred larger, more centralized states. Bates (forthcoming) shows that among the African


societies documented in the Standard Cross Cultural Sample (SCCS), there is a positiverelationship between rates of warfare and the degree of political centralization.

Other evidence suggests that a history of warfare can also have negative long-termimpacts. An innovative paper by Jancec (2012) provides evidence that interstate conflicthas a long-term adverse impact on trust in the political system today. Using individual-level data from Slovenia, Croatia, Serbia, Montenegro, Romania, and Ukraine, the authorshows that individuals living in regions that experienced more frequent changes in theruling nation-state between 1450 and 1945 have lower levels of trust in political insti-tutions today. In other words, populations whose ancestors more regularly experiencedconquest trust the government less today. In addition, Jancec (2012) also finds that fre-quent border change is associated with greater identification with an individual’s localityrather than with the nation, and with less participation in national politics, as measured byvoting. Since Jancec’s (2012) analysis always controls for broad region effects, he is able torule out worse governance as the explanation for the findings. All estimates are derivedholding national institutions constant. The most likely explanation is that a history ofconquest generated less identification with the nation-state and less trust for nationalleaders and institutions.

A theme that has developed in the literature highlights important historical linksbetween warfare and religion. For example,Botticini and Eckstein (2005,2007) argue thatthe burning of the Second JewishTemple by the Romans in 70AD had lasting importantconsequences for the Jewish religion and subsequent economic development. After theburning of the temple, the Jewish religion was transformed from one that centered aroundreligious sacrifices in the Temple to one that required all Jewish males to read the Torahand teach their sons to do the same in the synagog. According to Botticini and Eckstein(2005), the literacy and numeracy generated by this religious requirement resulted in themigration of Jewish farmers to cities,beginning in the fifth and sixth centuries,where theyengaged in urban occupations. In Babylon, Jews moved into urban centers and engagedin shopkeeping and artisanal activities such as tanning; linen and silk production; dyeing;and glassware-making. Jewish migration continued in the Muslim empire between themid-8th and early-9th centuries, with Jews entering a variety of skilled occupations,including handicrafts and jewelry production; ship building; money lending; and long-distance trading. Overall, their analysis provides compelling evidence that one importantviolent event—the Roman burning of the Jewish Temple—had impacts that affectedsubsequent human capital accumulation and the trajectory of economic prosperity of theJewish people.

Another example is the link between warfare with the Ottoman Empire and the rise ofProtestantism in Europe. Iyigun (2008) tests the established hypothesis that the Ottomanmilitary incursions into continental Europe from the mid-15th to late 16th centuriesallowed Protestantism to develop. Chronologically, a cursory examination of the datesof Ottoman and counter-reformation conflicts is consistent with the hypothesis. For

362 Nathan Nunn

example, the deadliest of all religious wars, the Thirty Years War (1618–1648), followedthe decline of the Ottoman Empire, marked by the Battle of Lepanto (1571) wherethe Holy League (a coalition of Catholic maritime states organized by Pope St. PiusV)defeated the Ottoman Empire’s primary naval fleet. Iyigun shows that the hypothesisreceives support when examined more rigorously. He analyzes annual data from 1450to 1700 and shows that in years with more European-Ottoman conflict there is lessconflict between countries within continental Europe, and there are less conflicts due toCatholic-Protestant religious differences.

Acemoglu et al. (2011a) examine the beneficial impacts of one of the most importantepisodes of European history: the French Revolution. Within a few short years (1789–1799), the revolution displaced traditional values regarding order and hierarchy with newenlightenment values of equality, citizenship, and inalienable rights. In 1792, the newrepublic declared war on Austria and its allies, including Prussia. Acemoglu et al. showthat the institutional reforms that were imposed on conquered territories had lastingimpacts in Germany and Prussia. Examining 19 regions at six different points in timebetween 1700 and 1900,the authors show that regions that experienced longer periods ofFrench occupation between 1793 and 1815, subsequently experienced faster economicdevelopment, measured by urbanization rates. Constructing an index of reforms thatquantifies the abolition of feudalism and guilds, and the implementation of the Frenchcivil code, they provide evidence consistent with these institutional reforms being thesource of the increased urbanization rates. Locations that experienced longer Frenchoccupation also had more intensive reforms.

7.3.4 Expulsions and Forced Population MovementsClosely related to warfare and conflict are expulsions and forced population movements.The most dramatic example of forced population movement is the export ofAfrican slavesduring the trans-Atlantic, trans-Saharan, Indian Ocean, and Red Sea slave trades. Slaveswere captured through kidnappings, raids, and warfare. Historical accounts suggest thatthe pervasive insecurity,violence,and warfare had detrimental impacts on state formation,inter- and intra-group co-operation, and institutional, social, and economic developmentgenerally (e.g. Inikori, 2000, 2003).

The most illustrative example of this is the experience of the Kongo Kingdom,whichwas discovered in 1493 by Diogo Cão. Initially, a diverse array of products were tradedbetween Kongo and the Portuguese, including copper, textiles, ivory, and slaves.The firstslaves that were traded were prisoners of war and criminals. However, the increasingexternal demand for slaves, the presence of Portuguese slave traders, and competition forthe throne within the Kingdom all resulted in a dramatic and uncontrollable increasein slave capture and raiding throughout the Kingdom. As early as 1514, King Afonsowas already writing to the Portuguese to complain of Portuguese merchants colludingwith noblemen to enslave Kongo citizens. In 1526,Afonso asked for the removal of all


Portuguese merchants and the end of trade. This attempt was unsuccessful and throughthe 16th century the process continued, culminating in the Jaga invasion of 1568–1570.The large-scale civil war from 1665 to 1709 resulted in the complete collapse of theKingdom (Heywood, 2009).

Nunn (2008a) empirically examines the impacts the slave trades had on the long-termdevelopment of the African continent. Combining information from historical shippingrecords with information from a variety of historical sources—plantation inventories,marriage records,death records,slave runaway notices,etc.—that report the ethnic identityof the slaves shipped from Africa, Nunn constructs estimates of the total number of slavesshipped from modern-day African countries during each of the four slave trades.

The study finds that the parts of Africa from which the largest number of slaves weretaken are the poorest today.The core issue in interpreting this correlation is selection intothe slave trades. If, for example, the societies with the most poorly functioning institutionsand the poorest future growth prospects selected into the slave trades, then this wouldexplain the negative relationship even if the external trade in slave trades had no directimpact on societies within Africa. Nunn tests whether selection is driving the resultsby looking at the evidence on the nature of selection during the slave trades. He findsthat the descriptive and quantitative evidence suggest that it was not the least developedsocieties that selected into the slave trade, but it was actually the more developed andmore densely populated societies that supplied the largest numbers of slaves. Nunn alsoconstructs instruments based on the distance of each country from the external locationsof demand for the slaves. He argues that although the location of the demand for slavesinfluenced the location of supply, the reverse was not true. The IV estimates provideestimates consistent with the OLS estimates. Nunn concludes that the empirical evidencesuggests that Africa’s external trade in slaves did have a significant negative impact on theeconomic development of regions within Africa.

Subsequent studies have documented other important impacts of Africa’s slave trades.Nunn and Wantchekon (2011) provide evidence that the slave trades adversely affectedsubsequent levels of trust within Africa. They show that the lower levels of trust arisethrough two channels: a deterioration of domestic institutions that enforce trustworthybehavior and an increase in the prevalence of cultural norms of distrust. They estimatethat quantitatively the second determinant is about twice as important as the former.8

Dalton and Leung (2011) and Fenske (2012) provide evidence that the trans-Atlanticslave trade resulted in a long-term increase in the prevalence of polygamy. This is due to

8 Deconinck and Verpoorten (forthcoming) update the results of Nunn and Wantchekon (2011) usingthe more recent (2008) round of the Afrobarometer survey. This increases the sample by two additionalcountries (from 17 to 19) and expands the number of ethnic groups from 185 to 228. With the morerecent and expanded sample, they find estimates that are very similar to Nunn and Wantchekon (2011).Also see Pierce and Snyder (2012) who show that in countries that were more impacted by the slavetrades,firms have less access to external financing today,whether it be through formal or informal means.

364 Nathan Nunn

the fact that primarily males were captured and shipped to the Americas, which resultedin a shortage of men within Africa. Interestingly, there is no evidence of such an impactfor the Indian Ocean slave trade, where there was not a strong preference for male slaves.Dalton and Leung (2011) conclude that Africa’s history of the slave trades is the primaryexplanation for why polygamy is much more prevalent inWest Africa than in East Africa,today.

Within Europe, studies have also found evidence of large persistent impacts of forcedmigration.Acemoglu et al. (2011b) examine the long-term impacts of the mass movementand murder of Jewish populations in Russia during World War II. Examining variationacross 278 cities, the authors show that Jewish depopulation during the holocaust isassociated with significantly slower population growth,which was still detectable 50 yearslater in 1989, the last year of their sample. The authors confirm these results by lookingacross 48 oblasts, identifying a relationship between Jewish depopulation and lower percapita income in 2002.

A number of studies have also examined the persistent impacts of the 1609 expulsion ofapproximately 300,000 Moriscos (Spanish Muslims) from the Iberian Peninsula. Chaney(2008) and Chaney and Hornbeck (2013) examine the effects in the Kingdom of Valencia,where 130,000 Muslims—equal to one-third of its total population—were expelled.Chaney and Hornbeck (2013) show that after the expulsion, total output respondedquickly although total population did not, resulting in higher per capita incomes indistricts where a greater share of the population had been expelled. The persistentlyhigher output per capita is potentially explained by the presence of more extractiveinstitutions with a higher tax rate that inhibited population growth. Chaney (2008) alsoexamines the impacts of the 1609 expulsion inValencia, but considers spillover impactson neighboring districts as low-skilled migrants moved to newly available land.

Forced movements of indigenous populations were also common in the Americas.Dippel (2011) examines the long-term development impacts of forced integration ofdifferent tribal bands onto the same reservation in the 19th century. He measures forcedintegration by combining information on the indigenous integration of the bands withina tribe (specifically, whether bands within a tribe were politically integrated prior to the19th century) with information on which bands were subsequently forced to live on thesame reservation. Forced integration occurs when bands that were previously indepen-dent were forced to live on the same reservation. He finds that reservations that expe-rienced forced integration have 30 percent lower per capita GDP in 2000. He providesconvincing evidence that this effect is causal and that it is due to dysfunctional politicalinstitutions.

Feir (2013) considers the impacts of the policy of forcibly removing indigenouschildren from their homes and sending them to residential schools. She finds that inCanada the schools were successful in their intended goal of eroding indigenous culture.


Individuals attending residential schools, as adults are 16 percent less likely to participatein traditional activities and 8 percent less likely to speak their indigenous language. Res-idential schools were notorious for the presence of mental, physical, and sexual abuse.Collecting data on the number of proven abuse claims by school, Feir shows that atten-dance in the least abusive schools is associated with increased educational attainment andmore employment. On the other hand, attending the most abusive residential schools isnot associated with increased education, but is associated with lower employment, lowerrates of marriage, and increased alcohol consumption.

7.3.5 ReligionA number of studies provide evidence of the persistent long-run impacts of importantreligious historical events.The episode that has received the most attention in the literatureis the Protestant Reformation,whose origin dates back to October 31,1517 when MartinLuther posted the Ninety-FiveTheses on the Power and Efficacy of Indulgences on the doors ofAll Saints’Church inWittenberg. He objected to corruption in the Catholic Church, andin particular, to the selling of indulgences. His teachings quickly spread, partly facilitatedby the recent invention of the printing press (Rubin, 2011).

According toWeber (1930), the new religion that emerged, Protestantism, was signif-icant because, in contrast to Catholicism, it approved the virtues of hard work and theaccumulation of wealth and that these values provided the moral foundation that spurredthe transition to a modern market-based industrial economy. Another significant featureof the Protestant religion is its emphasis on the ability of individuals to read the Bible.With this came a belief in the importance of education.

A large number of studies have examined the historical and persistent impacts of theProtestant religion. Becker andWoessmann (2009) examine the two potential impacts ofProtestantism,namely increased education and a change in values related to accumulation,thrift, and hard work.Their analysis examines variation in the intensity of Protestant andCatholic denominations across 452 counties in late 19th century Prussia. They find thatthe Protestant religion is associated with higher literacy.To better understand whether thecorrelations reflect a causal impact of the Protestant religion, they use a county’s distancefrom Wittenberg, the origins of the Reformation, as an instrument for the share of thepopulation that is Protestant in 1871.9 Using the same empirical structure,the authors alsoidentify a positive impact of Protestantism on various measures of economic development.This finding is consistent both with Protestantism increasing education which increasesincome, and with Protestantism affecting beliefs and values which increase income. Theauthors attempt to disentangle the two by estimating the impact of Protestantism onincome after netting out the level of income explained by education (which they estimate

9 The determinants and dynamics of the adoption of Protestantism are an interesting subject of analysis inits own right. For more on this, see Rubin (2011) and Cantoni (2012).

366 Nathan Nunn

directly and take from previous studies). The findings from this procedure indicate thatProtestantism’s positive impact on income can be almost fully explained by its impact oneducation.

The link between Protestantism and education also receives support from studiesexamining the long-term impacts of missionary activities outside of Europe. A largeliterature has emerged documenting this relationship in various locations and time peri-ods. An early contribution is provided by Woodberry (2004), who documents a positiverelationship between measures of the historical presence of missionaries and current percapita income and democracy across former non-settler colonies. According to Wood-berry these benefits arise not only from increased education, but because missionaries,particularly Protestant missionaries, fought against injustices against native populationsduring colonial rule, which helped to foster better institutions, improved civil liberties,and increased democracy in the long-run (Woodberry, 2004, 2012).

Others have also examined the impact of missionary activities, but use a more micro-approach that focuses on a specific region or country. For example,Bai and Kung (2011b)look within China and examine county-level data from 1840 and 1920. They identify apositive relationship between Protestant missionary activity and economic development,measured using urbanization rates.

A recent insight within this literature is the identification of differences betweenreligious denomination or orders within the Protestant and Catholic religions.Waldinger(2012) examines variation within colonial Mexico and shows differences in the long-termimpacts of four different Catholic orders: the Franciscans, Dominicans,Augustinians, andJesuits. She finds that the three Mendicant orders (Franciscan, Dominican,Augustinian),which shared a strong commitment to alleviating poverty and educating the poor, had along-term impact on educational attainment. By contrast, the Jesuits, who focused theireducational efforts on the colonial elites only, appear to have had long-term effects onconversion to Catholicism, but not on increased educational attainment.

Andersen et al. (2011) analyze the Catholic Order of Cisterians in England during theearly modern period. One defining characteristic of the Catholic order,which after beingestablished in France in 1098,quickly spread across England in the following century,wasthe belief and emphasis on a strong work ethic and promotion of thrift. Examiningcounty-level data for England from 1377 to 1801, the authors show that counties witha greater presence of Cisterian monasteries exhibited greater population growth duringthis period.

Akçomak et al. (2012) empirically trace the impacts of the founding of the Brethrenof the Common Life, a Roman Catholic community established by Geert Groote inthe late 14th century. The movement arose because of dissatisfaction with the CatholicChurch and set to reform the Church by educating citizens and enabling them to read theBible in the vernacular. In addition to their strong emphasis on literacy and education,


the Brethren of the Common Life also promoted hard work and productive labor.10

The authors empirically trace the historical impact of the Brethren of the Common Lifewithin the Netherlands. Examining a sample of Dutch cities, the authors show that citieswith Brethren of the Common Life communities had higher rates of literacy in 1600,more book production from 1470 to 1500, and faster population growth between 1400and 1560. Of course, these correlations may be driven by reverse causality or omittedvariables bias. The authors attempt to better understand whether the correlations arecausal by using a city’s distance from Deventer, the birthplace of Geert Groote and theorigins of the movement.

Gender differences between the Protestant and Catholic religions are another aspectthat has been examined by the literature. Because Protestants believed that reading theBible directly was important for salvation, even for women, they placed greater impor-tance on female education than Catholics. Using data from the first Prussian census of1816, Becker andWoessmann (2008) show that Protestantism is associated with a smallergender gap in education. Evidence for a greater emphasis on female education amongProtestants is also found in Nunn’s (forthcoming) analysis of the impacts of Catholicand Protestant colonial African missions on long-term education. He finds that althoughboth had positive impacts on long-term education, the impact of Protestant missions wasconcentrated among females, while the impact of Catholic missions was concentratedamong males.

7.3.6 Technological InnovationFindings from a number of recent studies suggest a link between innovative activities inthe past and subsequent economic outcomes. For example,Comin et al. (2010) documenta positive correlation between the measure of a society’s level of technology in the past(either 1000 BC,0AD,or 1500AD) and either its level of technology or per capita incometoday. The authors hypothesize that this is driven by increasing returns to technologyadoption: a higher level of technology lowers the cost of discovering new technologies.That is,a higher level of technology in the past affects the ease of accumulating subsequenttechnology which impacts technology in the future. Of course, their findings are alsoconsistent with omitted persistent factors impacting both technology and developmentin the past and today.An example is the persistence of governance and institutional qualityas has been documented by Bockstette et al. (2002).

Other studies, by zooming in on specific innovations, have been more successfulat establishing persistent long-term impacts. Dittmar (2011) examines the long-termeffects of the printing press, which was first established in Mainz, Germany between1446 and 1450. He constructs a panel of European city-level data at 100-year intervals

10 The similarity between Protestant beliefs and the Brethren of the Common Life is not a coincidence, asMartin Luther studied under the Brethren of the Common Life at Magdeburg before attending university.

368 Nathan Nunn

between 1300 and 1800, combining data on city populations with information on theearly adoption of the printing press. His analysis shows that cities that adopted the printingpress between 1450 and 1500 experienced faster population growth during the 16th to19th centuries.The impacts he estimates are extremely large.They imply that the printingpress accounts for 18% of city growth between 1500 and 1600.

Dittmar uses the panel dimension of his data to validate his cross-sectional finding,showing that cities that adopted the printing press in the late 15th century were notgrowing more quickly prior to adoption. This suggests that the results are not drivenby unobserved time-invariant differences between cities. He also provides additionalevidence for a causal interpretation of his estimates using distance from the inventionof the printing press—Mainz, Germany—as an instrument for adoption in the late 15thcentury. The IV estimates are consistent with the OLS estimates.

In a follow-up study, Dittmar (2012) calculates the impact of the printing press onaggregate welfare. Using data from England on the price and consumption of printedbooks in England between the 1490s and 1700 (and assumptions about consumers’utilityfunctions), he estimates that the printed book increased welfare by an equivalent of 4%of income; by the mid-17th century this figure was 3–7%.

Baten and van Zanden (2008) provide complementary evidence of the importanceof the printed book for long-term growth. The authors construct an impressive datasetof the production of printed books in eight Western European countries every 50 yearsbetween 1450 and 1800. The authors show that book production correlates stronglywith literacy, and in panel regressions with time-period fixed effects, initial per capitabook production is positively associated with faster growth in real wages during the next50 years.

Alesina et al. (2013) examine the long-term impacts of the plough, an importanttechnological innovation used in agriculture. The tool, which was able to prepare largeamounts of soil for cultivation in a shorter period of time than previous tools, was firstinvented between 6000 and 4000 BC in Mesopotomia (Lal et al. 2007). Although theimpacts of the plough were likely vast, the authors focus on one consequence that washighlighted by Boserup (1970). Because the use of the plough required significant upperbody strength, it tended to generate a gender division of labor where men worked outsidethe home in the fields while women specialized in home production and other domesticactivities. Boserup argues that this gender division of labor resulted in deeply held beliefsabout the role of women in society. In societies that traditionally use plough agriculture,less equal beliefs about the roles of men and women evolved. Alesina et al. (2013) test thishypothesis by linking ethnographic data with contemporary individual- and country-levelmeasures of gender role attitudes. They find that traditional plough agriculture is associ-ated, even today, with less equal beliefs about the roles of men and women in society.11

11 Also see Hansen et al. (2012), who find that a longer history of agriculture is associated with moreunequal gender roles.


The findings of Alesina et al. are consistent with evidence that in the early phase ofagriculture, prior to the adoption of the plough, societies tended to be matriarchal andcharacterized by gender equality (Gimbutas,2007). Recent excavations from Çatalhöyük,a Neolithic town of 8000 people on the plains of centralTurkey inhabited approximately9000 years ago, provide additional evidence of the equality of the sexes during this time(Hodder,2005).Analysis of male and female skeletal remains shows carbon deposits insidethe ribs, due to indoor wood fires and a lack of ventilation in the homes. The smokewas ingested causing soot to build up in the lungs resulting in a lining of carbon insidethe ribs. Hodder (2005) finds that the average amount of carbon in the ribs of menand women was equal, suggesting that men and women tended to spend roughly equalamounts of time within and outside the home. In addition, the archeological evidencefrom Çatalhöyük suggests that men and women had similar diets and were buried insimilar positions and locations, both of which also suggest roughly equal social status.

The growth-promoting impacts of the plough have been studied by Andersen et al.(2013). Examining 316 European regions between 500 and 1300AD, the authors showthat the adoption of the heavy plough is associated with greater population growth andincreased urbanization. According to the authors’ diff-in-diff estimates, the heavy ploughaccounts for 10 percent of the increase in population and urbanization during this time.

7.4. GEOGRAPHY ANDHISTORY

7.4.1 The Historical Impacts of GeographyOne of the important insights that has arisen from the historical development literatureconcerns the relationship between geography and contemporary development. Specifi-cally, a common finding in the literature is that geography can have important impactson current development through its persistent historical effects. Further, evidence alsosuggests that this historical impact of geography may be larger than its contemporaneousimpact. For example, the findings from Acemoglu et al. (2001) show that the diseaseenvironment at the time of European colonization crucially affected subsequent insti-tutional development. The authors argue that the primary impact of a country’s diseaseenvironment works through this historical channel rather than through contemporarychannels. The line of research by Engerman and Sokoloff (1997, 2002) also shows thatsmall geographic differences become magnified through historical events and as a resultend up having large impacts on long-term economic development. As they argue, dif-ferences in soil and climate made plantation agriculture and its reliance on slavery moreor less profitable in different parts of the Americas, which in turn affected long-termeconomic development.

Even more dramatic examples of the long-term historical effect of geography aredocumented by Jared Diamond in his book Guns, Germs and Steel. The book is devotedto exploring the answer to the question of why Europeans colonized the rest of the world

370 Nathan Nunn

and not the other way around. Part of Diamond’s answer lies in the fact that in Eurasia,crops and animals were domesticated earlier and in more varieties than in other parts ofthe world.

In addition, the domestication of plants and animals quickly spread east and westthroughout Eurasia, but diffused much less quickly south to the African continent.Whenmoving east or west, the length of the day does not change, and the climate is generallynot drastically different. However, this is not true when moving north or south,where thelength of the day changes and the climate typically is very different. More generally, forcontinents with a north-south orientation, such as the Americas or Africa, domesticationor technological advance tended not to spread as quickly as in Eurasia with its east-westorientation.

Because of the early domestication of animals in Eurasia (and its more rapid diffusion),humans lived in close proximity to animals. As a result of this, new animal-based diseases,such as measles, tuberculosis, influenza, and smallpox developed, and over time humansdeveloped genetic resistances to the diseases. In contrast, the parts of the world withoutdomesticated animals did not develop the diseases or genetic resistance. According toDiamond, this explains why European diseases decimated native populations and not theother way around. As Diamond points out, the spread of disease was as important a factoras the military for European conquest of the Americas.

Diamond’s explanation for Europe’s global dominance illustrates clearly the large effectthat geography can have through history.The historical origins of European colonizationof the globe lie in two deep determinants: (i) being endowed with wild plants andanimals suitable for domestication; and (ii) being located on a continent with an east-west orientation.

Although Diamond’s hypothesis is intuitive in many ways, there are reasons to besceptical. First, having domesticated plants and animals is potentially endogenous. Forexample, Diamond asserts that although the horse was domesticable, its close relativethe zebra was not (Diamond, 2002). However, this assertion is difficult to assess sincewe do not observe the wild ancestors of the horse and so cannot compare it to thezebra. All we observe is the domesticated version, which has undergone centuries ofselective breeding. Perhaps there are other historical determinants—be they economic,cultural, institutional, etc.—that caused horses to become domesticated in Eurasia butnot zebras in Africa. Interestingly, there are examples of Europeans attempting to tamezebras. Rosendo Ribeiro, a doctor in Kenya, made house calls on a zebra. In England,LordWalter Rothschild, pictured in Figure 7.2, would frequently drive a carriage pulledby zebras through the streets of London. However, despite these examples, the zebranever become widely domesticated.

Olsson and Hibbs (2005) take Diamond’s hypothesis to the data. Using modern coun-tries as the unit of analysis, the authors show that consistent with Diamond’s descriptiveaccounts, countries with richer biological and geographic environments experienced the


Figure 7.2 Lord Lionel Walter Rothschild with his zebra-drawn carriage, 1895. source: The PictureMagazine.

transition to agriculture at an earlier date and have higher per capita GDP in 1997. Thegeographic environment is measured using an index that includes the axis orientationof the continent, suitability of the climate for agriculture, latitude, and the size of thelandmass within which the country is located. The measure of biological conditions isbased on an index that comprises the number of annual or perennial wild grasses knownto exist in prehistory and with a mean kernel weight exceeding 10 mg, as well as thenumber of domesticable mammals known to exist in prehistory and weighing more than45 kg. Overall, the authors find that their estimates confirm Diamond’s hypotheses.

Evidence from a wide range of empirical studies provides additional evidence ofthe importance of historical impacts of geography. Ashraf and Michalopoulos (2011)provide evidence that geography was an important determinant of the timing of theNeolithic Revolution, arguably the most important event in human history. Lookingacross countries globally, and across archaeological sites within Europe and the MiddleEast, they document an inverted U-shaped relationship between year-to-year variabilityin temperature and early adoption of agriculture.12

Michalopoulos (2012) shows that geography was an important determinant of theevolution of ethnic identity and hence ethnic diversity (which is known to be highlycorrelated with economic development today). Michalopoulos (2012) provides evidence

12 Because fine-grained temperature data are not available prior to 1500, the authors are forced to usepost-1500 variability as a proxy. The assumption is that the rank ordering of variability after 1500 issimilar to the ordering prior to 1500.They show that this is true comparing data from 1500 to 1900 and1900 to 2000.

372 Nathan Nunn

that the pattern of agricultural suitability and terrain slope were important determinantsof the interaction between ethnic groups and their proclivity to merge into and identify aslarger ethnicities. His analysis combines fine-grained geographic data with informationon the locations of ethnic groups globally. The world is then divided into grid-cellsthat are 2.5◦ by 2.5◦. Michalopoulos (2012) shows that grid-cells that exhibit greatervariation in soil quality and in elevation are also more linguistically diverse. The mostlikely explanation for the finding is that greater geographic variation prevented trade andmigration between societies, and conquest of one society over others, all of which havehomogenizing impacts.

Interestingly, Michalopoulos (2012) shows that the link between geography and eth-nic diversity is due to geography’s impact prior to 1500. Among the parts of the worldthat witness significant population changes after 1500 (due to death and voluntary andinvoluntary migrations), there is no relationship between geographic diversity and lin-guistic diversity.

Durante (2010) provides evidence that within Europe,historical variability in weatherconditions created greater benefits for cooperation, which increased the level of cooper-ation in societies. He hypothesizes that greater spatial variability in temperature and pre-cipitation generates output shocks that are less correlated,providing an increased incentivefor trade, thus increasing trust and cooperation. As well, greater temporal variability ofweather increases the benefits to large storage facilities and irrigation, which requirelarge-scale cooperation. Durante therefore argues that locations characterized by greaterspatial and temporal variability may have higher levels of trust and cooperation today. Hetests these predictions using monthly historical climatic data from 1500 to 2000, mea-sured across grid-cells within Europe. He finds that greater year-to-year variability inboth temperature and precipitation is associated with higher levels of trust today, and thatless correlated weather shocks over space is also associated with more trust today.

Of course, there are a number of potential alternative explanations for these correla-tions. Therefore, as a further test of his channel, Durante measures variability in growingseason months and months outside of the growing season. He finds that only historicalvariability during the growing season is correlated with current trust. He also examinesweather variability from 1500 to 1750,which was prior to the industrial revolution whenEurope was primarily agricultural, and from 1900 to 2000, which is after industrializa-tion. He finds that only weather variability during the agricultural period is correlatedwith trust today.13

Recent findings fromAlsan (2012) suggests that geography also had important histori-cal impacts within Africa. A large literature attributes many of Africa’s unique

13 A subsequent study by Ager and Ciccone (2012) raises the questions of whether the increased trustfound in Durante (2010) is due, in part, to increased religiosity. Although in a different context—the19th century United States—Ager and Ciccone (2012) show that increased variability in annual rainfall(looking across counties) is associated with increased church membership.


characteristics to the fact that it is land abundant and labor scarce. Alsan (2012) con-siders a potential explanation for this: the tsetse fly. The fly, which is unique to Africa,transmits the parasite trypanosomiasis,which causes sleeping sickness in humans and naganain domesticated animals. The tsetse fly, both directly through its impact on humans, andindirectly through its impact on domesticated animals,may be responsible forAfrica’s lowpopulation densities historically.

The author uses 19th century climate data measured at the grid-cell level to construct ameasure of the historical suitability of each cell for the tsetse fly.The index is a highly non-linear function of temperature and humidity. Examining variation across ethnic groupswithin Africa, she shows that ethnicities with climates more suitable for the tsetse fly, atthe end of the 19th century, were less likely to use draft animals for trade and agriculture,were less likely to use the plough, and were more likely to use shifting cultivation ratherthan more intensive agricultural techniques. Because tsetse-suitable areas did not developplough agriculture, women were more likely to participate in agriculture, and becausethe use of animals was not feasible, slaves were more likely to be used. Additionally, theless intensive agricultural techniques resulted in lower population densities, fewer urbancenters, and less developed states. Her findings provide strong evidence that geographicsuitability for the tsetse fly had a formative impact on the nature and prosperity of societieswithin Africa.

Because the tsetse fly did not exist outside of Africa, Alsan is able to undertake afalsification test by examining the correlation between tsetse suitability and the outcomeof interest in the other parts of the world. If her estimates are really capturing the causalimpact of the tsetse fly on long-term development, then in the parts of the world wherethere was no fly, we should not observe the same correlations. This is indeed what shefinds. The tsetse suitability index has no predictive power outside of Africa. Overall, herfindings provide strong evidence that the tsetse fly, by inhibiting the development ofintensive agriculture using draft animals, resulted in lower populations, less urbanization,and less state development.

Fenske (2011) also considers the question of how geographic conditions affected thehistory of state development in Africa. The author tests the hypothesis that ecologicaldiversity, by increasing the benefits of peaceful exchange between locations, increased theneed of a state to provide the institutional setting to facilitate trade. This in turn resultedin the development of larger more developed states. Combining data on the boundariesof African ethnic groups in the 19th century with information on 18 ecological zoneswithin Africa, Fenske constructs a measure of each ethnic group’s ecological diversity.He finds that ethnic groups that were more ecologically diverse also had larger and moredeveloped states.

A large number of studies also examine historical weather shocks and show that theyhad important historical impacts, many of which continue to be felt today. For example,Fenske and Kala (2013) show that during the slave trade,cooler temperatures near the slave

374 Nathan Nunn

ports were associated with increased slave exports.Therefore,due to the persistent impactsof the slave trade,temperature fluctuations during this time period had long-term impacts.Bai and Kung (2011a) examine the impacts of rainfall on Sino-Nomadic attacks in HanChina between 220BCE and 1839CE. They identify a negative relationship betweenconflict and precipitation showing that climate was also an important determinant ofconflict in the region.

Chaney (forthcoming) shows that in Ancient Egypt deviant Nile floods had impor-tant politcial impacts. Because deviant floods increased social unrest, this increased thebargaining power of the religious leaders relative to military leaders. Consistent with this,Chaney (forthcoming) shows that from 641 to 1437CE,deviant floods are associated withhigher food prices, more conflict, less turnover of the highest ranking religious leader,and more construction of religious structures (relative to secular ones).

Haber and Menaldo (2010) also argue that climate can have important political effects.They show that there exists an inverted U-shaped relationship between average rainfall anddemocracy. They argue that this relationship is explained by the non-linear relationshipbetween rainfall and suitability of a location for sedentary agriculture, which they argueprovides a foundation more suitable for democracy than nomadic modes of subsistence.Bentzen et al. (2012) also argue for a link between geography/climate and modernpolitical institutions, but, motivated by Wittfogel (1957), focus on the extent to whicha location’s agricultural output is increased by investments in irrigation. They arguethat the large-scale investment and coordination needed for irrigation promoted strongauthoritarian leadership and autocratic institutions, and this has persistent impacts eventoday. Using data from the FAO on yields with and without irrigation, the authorsconstruct a measure of irrigation potential. Looking across 160 countries,they find greaterirrigation potential is associated with less democracy today. Bentzen et al. (2012) showthat in their specification the non-linear effect found in Haber and Menaldo (2010) nolonger exists once one controls for their measure of irrigation potential.

Overall, there is a large body of evidence—only some of which is reviewed here—thatsuggests that a significant effect of geography—if not the largest effect of geography—on current economic development arises due to its influence on past events rather thanthrough its direct effect on economic outcomes today.

7.4.2 Geography’s Changing Impact Over Time and SpaceOnce one recognizes the fact that geography had important impacts historically as wellas today, it is natural to ask whether the impacts of geography have been roughly constantthroughout time or whether its impact varies in a systematic manner across time and/orspace. This is a point also addressed in Acemoglu et al. (2001). Their empirical andhistorical narrative is that the disease environment generally, and in particular today,does not have large impacts on economic development. However, during the period of


European colonization of much of the globe it had a crucial impact. In locations with adisease environment that threatened European survival, Europeans did not migrate anddid not establish growth-promoting institutions. Acemoglu et al.’s (2001) assumptionthat this disease environment only mattered during this specific historical episode is whatallows them to use initial settler mortality as an instrument for a country’s domesticinstitutions in explaining current per capita income. According to them, this particulargeographic characteristic—the severity of the disease environment for Europeans—onlyhad impacts during the colonial period.

Along similar lines, a number of papers find evidence that weather shocks can havesignificant long-term impacts during specific windows of time and no long-term effectsduring others. For example,Dell (2012) shows that within Mexico, drought experiencedby municipalities between 1906 and 1910 had a large positive impact on violence andinsurgency during the Mexican Revolution (1910–1918),resulting in a greater prevalenceof ejidos (communal farms), which are less economically developed today. This impliesthat drought experienced between 1906 and 1910 had a long-term persistent impact onunderdevelopment in Mexico. She shows that drought in other periods (between 1960and 1995) are uncorrelated with long-term development.

Osafo-Kwaako (2012) also finds evidence of weather shocks mattering during a spe-cific window of time. He shows that withinTanzania and during the early process of thegovernment’s establishment of development villages in the early 1970s, drought provideda motivation for peasants to agree to the villagization process. One therefore observes apositive relationship between droughts in 1973–1975 and the subsequent extent of vil-lagization. The author then documents the persistent impacts of villagization. Althoughit increased education levels, political awareness, and community participation, it has alsoled to increased poverty and lower consumption today. Like Dell, Osafo-Kwaako alsoshows that the long-term impacts of drought are specific to this one narrow window.

Fenske and Kala (2013), in their study of the link between climate and slave exportsin 18th and 19th century Africa, also provide some suggestive evidence of climate beingparticularly important during the height of the slave trade. They estimate the cross-sectional relationship between contemporary light density (a commonly used measureof economic development at the sub-national level) and historical weather shocks.Theirfindings provide evidence of the greater importance of temperature shocks during theheight of the trans-Atlantic slave trade,which is consistent with the shocks having a largeimpact on contemporary development through their historical impacts on the supply ofslaves.

Nunn and Puga (2012) focus on geography and provide an example of its impactvarying over both time and space.They show that for most of the world,terrain ruggednesshas a negative contemporaneous impact on economic development.All else equal, ruggedterrain makes it more difficult to build buildings, roads, bridges, and other infrastructure;

376 Nathan Nunn

agriculture and irrigation is also more difficult; and trade is more costly. They furthershow that within Africa, ruggedness had very different impacts than outside. WithinAfrica, greater ruggedness is associated with higher incomes, not lower.

The authors argue and provide evidence that this can be explained by an indirecthistorical impact of geography that was specific to Africa because of its history of theslave trades. During the slave trades, societies were able to use rugged terrain to protectand hide from slave raiders and kidnappers.This allowed individuals,villages, and societiesto partially defend against the negative effects of the slave trades documented in Nunn(2008a). Therefore, for the African continent, which was exposed to the slave trade,ruggedness also had a historical indirect positive effect on income. Ruggedness allowedcertain areas to evade the slave trade, thereby increasing long-term economic growth.

Nunn and Qian’s (2011) study of the introduction of the potato to the Old Worldduring the Columbian Exchange directly exploits the fact that the importance of geog-raphy changes over time. Specifically, their analysis relies on the fact that having climateand soil suitable for cultivating potatoes was important only after the potato was intro-duced from the Americas. Despite not having spatially or temporally extensive data onpotato production or consumption, they infer the impacts of the potato by comparing theevolution of populations, city sizes, urbanization rates, and adult heights, before and afterthe adoption of the potato, in the places suitable for potato cultivation relative to unsuit-able locations. Their estimates show that after the introduction of the potato, the placessuitable for cultivation witnessed significant population growth, city growth, increasedurbanization rates, and increased heights.

Overall, evidence continues to accumulate suggesting that geography can have verydifferent impacts at different points in time and in different locations. The impacts ofgeography depend crucially on the particular historical context.

7.5. MECHANISMS UNDERLYING HISTORICAL PERSISTENCE

I next turn to the important question of why historical events have persistentimpacts. In particular, I discuss the existing evidence for path dependence, culture, insti-tutions, and genetic traits as important channels underlying historical persistence.

7.5.1 Multiple Equilibria and Path DependenceAlthough it is far from obvious why historical events have persistent impacts,particularly inthe long-run,once one acknowledges the possibility of multiple equilibria, then historicalevents can have long-term impacts if they move the society from one equilibrium toanother. A large number of models show how easily multiple equilibria arise, even invery simple environments. See, for example, Murphy et al. (1993), Acemoglu (1995),Mehlum et al. (2003), and Nunn (2007).


Less formally, many examples of multiple equilibria in daily life have been identi-fied, the most well known being the adoption of the less-efficient QWERTY keyboardover other more efficient configurations like the DVORAK keyboard (David, 1985).The QWERTY keyboard design was developed by Christopher Sholes and patented in1873. That same year, it was sold to Remington, which used the configuration for theirtypewriters.The configuration was chosen because it separated the most commonly usedkeys, which kept the arms of the typewriter from jamming. In other words, the formatwas chosen because it effectively reduced typing speeds.14

A number of studies have undertaken the task of formally testing for the existence ofmultiple equilibria.A common strategy that has been employed is to examine cases wherethere has been an extremely large temporary shock to an equilibrium. The studies thentest whether the temporary shock causes a permanent movement to a new equilibrium.If so, this is evidence for the existence of multiple equilibria.

Davis and Weinstein (2002, 2008) examine the effect of bombings on 114 Japanesecities during World War II and show that after the bombings, the cities returned to theirpre-bombing populations, regained their shares in total manufacturing output, and mostsurprisingly, also regained their pre-existing industrial composition. Overall, the resultspoint toward the existence of a unique stable equilibrium of production, rather than theexistence of multiple equilibria.

Although these results appear to suggest the existence of one unique equilibrium, asecond possibility is that the shock was not sufficient to move the society away fromthe current equilibrium.The US bombings duringWWII were dramatic and severe, butthey did not alter property rights or the ownership of assets. It is likely that these are thefundamental determinants of where people live and where production occurs.

The findings in Miguel and Roland’s (2011) analysis of the long-term effects of the USbombings in Vietnam are consistent with the finding from Davis and Weinstein (2002,2008). The authors find that the bombings had no long-term effects on populations,poverty, or consumption 25 years later. However, in this case, the authors show that thereturn can be explained by reconstruction efforts intentionally aimed at rebuilding thehardest hit parts of the country. In other words, policy intentionally helped the countryreturn to its original equilibrium.

An innovative study by Redding et al. (2011) tests for the existence of multipleequilibria in a very different setting. The study examines the location of airport hubsin Germany before and after the division of Germany following World War II. It isshown that after division, the location of West Germany’s primary airport hub switched

14 Liebowitz and Margolis (1990) argue that the efficiency difference between the QWERTY and DVO-RAK keyboards is lower than argued in David (1985). The authors provide some evidence for this.However, even if the efficiency gap is lower than previously thought as they contend, the QWERTYkeyboard still provides a clear example of multiple equilibria and path dependence, which is the centralpoint of David’s (1985) original argument.

378 Nathan Nunn

from Berlin to Frankfurt. After reunification in 1990, the location of the hub did notswitch back to Berlin. Redding et al. show that this shift cannot be explained by changesin fundamentals over the time period. Thus, the evidence suggests that the temporarydivision of Germany resulted in a permanent movement of the location of Germany’slargest airport hub.

Bleakely and Lin (2012) examine a very specific and seemingly innocuous geographiccharacteristic and show that even though it only mattered for a narrow window of time, ithad lasting and important impacts on urban development within the United States. Thecharacteristic they examine is the existence of rapids or falls, which occur when a rivercrosses a fault line. In these locations, river transport required hauling goods and boatsover land. This is known as portage. These locations were a focal point for commercialactivity and entrepot trade.

The shipment of goods by boat was a dominant form of transportation until the early tomid-19th century,when canals and railways were developed. Combining geographic datawith population at the census tract level, the authors show that today, looking either alongrivers or along fault lines, populations are concentrated where rivers cross fault lines—i.e.at historical portage sites.The authors then turn to historical populations, examining therelationship between portage and population density from 1790 to 2000. The authorsshow that after 1850 (and the decline in the use of water transport and portage), thepopulation actually became more (not less) concentrated at portage sites. Their findingsare consistent with portage sites serving as a focal point that helped determine the locationof early cities (i.e. the equilibrium population distribution) among a large set of possiblemultiple equilibria.

7.5.2 Domestic InstitutionsEven without the existence of multiple equilibria, historical events can still affect eco-nomic development in the long-run if they alter deep determinants of long-term eco-nomic growth.The determinant that has received the greatest attention in the literature isdomestic institutions.This focus is illustrated by the fact that in each of the seminal papersby Acemoglu et al. (2001,2002),Engerman and Sokoloff (1997,2002), and La Porta et al.(1997, 1998), the mechanism through which colonial rule affects current development isinstitutions.

The focus on institutions as a causal mechanism has also continued in subsequentresearch. An example of this is Acemoglu et al.’s (2005) study of the effect that earlyAtlantic trade had in Europe. The authors argue that in countries with access to thelucrative Atlantic three-corner trade, economic and political power shifted toward com-mercial interests. As the merchant class became more powerful, they were able to alterdomestic institutions to protect their interests against the interests of the royalty, and theseinstitutional changes in turn had a positive effect on long-term prosperity. Using data onhistorical urbanization rates and per capita incomes, the study first shows that the rise of


Europe was actually a rise of nations with access to the lucrative Atlantic trade, namelyBritain, France, the Netherlands, Portugal, and Spain.

The authors argue that profits alone are not able to explain the divergent growth ofAtlantic traders and that the evolution of domestic institutions played an important role inthe process.15 To test this hypothesis, the authors extend the Polity IV data back to 1350and show that Atlantic trade increased the quality of domestic institutions as measured byan index of the constraints on the executive. They further hypothesize that the processof institutional change could only occur in countries that initially had non-absolutistpolitical institutions.They show that the data are also consistent with this.The increase ineconomic growth generated by Atlantic trade was higher for countries with better initialdomestic institutions, again measured by the constraint on the executive.

Other examples of studies documenting the persistent importance of historical insti-tutions include Dell’s (2010) analysis of the impact of the early forced labor institutionsin colonial Peru and Bolivia, as well as Banerjee and Iyer’s (2005, 2008) studies of theeffects of early land tenure institutions in colonial India.

The recent study by Gennaioli and Rainer (2007) also provides evidence of the persis-tence of early institutions, but within the African context.The authors use ethnographicdata to construct a measure of the level of state development in pre-colonial Africansocieties. Their OLS estimates show that there is a positive correlation between pre-colonial political development and the provision of public goods today. More recently,Michalopoulos and Pappaioannou (2013) combine the same ethnographic data used inGennaioli and Rainer (2007) with satellite data on night-light density. Examining within-country variation, the authors find that the only robust correlate of night-light density isan ethnicity’s pre-colonial level of political development. This finding echoes Gennaioliand Rainer’s finding of the importance of this variable.

These results can be combined with evidence from Nunn (2008a) showing that theparts of Africa from which more slaves were taken had less developed political systemsafter the slave trade (and before official colonial rule).16 Although the evidence for bothrelationships is based on correlations and therefore one must be cautious when drawingconclusions, the combined evidence from Gennaioli and Rainer (2007), Michalopoulosand Pappaioannou (2013), and Nunn (2008a) is consistent with a chain of causality wherethe slave trade resulted in a deterioration of domestic political institutions, which in turnhad a long-term adverse impact on the provision of public goods.Therefore, the body ofevidence provides support for the notion that history can matter through the evolutionand persistence of early institutions.

Overall, the literature since Acemoglu et al. (2001) has succeeded at providing addi-tional evidence showing that institutions are an important channel through which history

15 See Inikori (2002) for the alternative view that the profits that accrued by Western Europe during thethree-corner Atlantic trade explain much of its growth during that time.

16 Also see Whatley (forthcoming) for micro-level evidence for this relationship.

380 Nathan Nunn

matters. However, much work remains to be done before we have a clear understand-ing of the effect that historical events have on the formation of early institutions andtheir persistence and importance for long-term development. For example, in past stud-ies (typically at the macro-level) institutions have been conceptualized and measured asa broad cluster of institutions. The result of this is that, by-and-large, institutions haveremained a black box that we do not clearly understand the details of.17 As empiricalresearch continues to examine specific examples of institutional change and persistence atthe micro-level,our understanding of the causes and consequences of specific institutionswill naturally improve.

7.5.3 Cultural Norms of BehaviorAnother way in which historical events can have long-term impacts is if these past eventspermanently affect culture or norms of behavior. While in economics the notion ofculture often remains vague, other disciplines place much more emphasis on preciselydefining culture. For example, evolutionary anthropologists have long recognized thatthere are clear micro-foundations that explain the existence of a phenomenon like cul-ture (e.g. Cavalli-Sforza and Feldman, 1981; Boyd and Richerson, 1985). If informationacquisition is either imperfect or costly, then selection favors short-cuts to learning. Indi-viduals, rather than using scarce resources to acquire all of the information needed forevery decision to be made, will instead develop “rules-of-thumb”.These short-cuts thenbecome internalized as individuals come to believe that certain behaviors are the “right”behaviors in certain situations.18 For a fuller exposition of this definition of culture seeNunn (2012).

The idea that norms of behavior may be a channel through which history can affectlong-term economic development is not new. One of the most famous links betweenhistory, culture, and development is Max Weber’s (1930) hypothesis that the ProtestantReformation was instrumental in facilitating the rise of industrial capitalism in WesternEurope. He argues that Protestantism, in contrast to Catholicism, approved the virtuesof hard work and the accumulation of wealth, and that these values, referred to as the

17 An exception is the study byAcemoglu and Johnson (2004),where the authors distinguish“property rightsinstitutions” from “contracting institutions.” According to their definitions, property rights institutionsprotect individuals from theft or expropriation by the government or elites, and contracting institutionsenforce private contracts written between individuals. They find that property rights institutions havea positive and significant effect on income, investment, and financial development. On the other hand,contracting institutions appear to have a much more limited impact, only affecting the form of financialintermediation.

18 Within economics, examples of models of cultural evolution includeVerdier (2000, 2001) and Tabellini(2008).


“Protestant work ethic,” provided the moral foundation that spurred the transition to amodern market-based industrial economy.19

One of the earliest studies empirically examining the possibility that cultural normsmay be historically determined was undertaken by a group of social psychologists (Cohenet al. 1996).The authors test whether there is a culture of honor in the US south,where aspecial importance is placed in defending one’s reputation and honor, even if this requiresaggression and violence.Their explanation for why this culture exists in the US south andnot the north lies in the different histories of settlement in the two areas. The north wassettled by groups with a farming background,while the south was settled primarily by theCelts who had been herders since prehistorical times and had never engaged in large-scaleagriculture. They argue that in herding cultures, with their low population densities andweak states, protection of one’s property was left to the individual and therefore normsof aggressive behavior developed as a means to protect one’s herd.

To test the culture of honor hypothesis, Cohen et al. (1996) conducted a series ofexperiments involving white males from the US north and US south. In the experiments,each individual was bumped by an accomplice and called an “asshole.” (The participantsdid not know this was part of the experiment.) Cohen et al. use a number of methodsincluding direct observation, psychological tests, and saliva tests to compare the effects ofthis incident on southerners relative to northerners. They find that southerners becamemore upset, were more likely to feel that their masculinity was threatened, became morephysiologically and cognitively primed for aggression as measured by a rise in testosteroneand cortisol levels, and were more likely to engage in aggressive behavior, subsequently.

A number of studies provide additional evidence for the historical origins of currentcultural differences. For example,Guiso et al. (2008) empirically examine the well-knownhypothesis put forth by Putnam et al. (1993) that within Italy, city states that becameindependent during the 1000–1300 period developed higher levels of social capital, andthese higher levels of social capital continue to persist today. The authors bring Putnamet al.’s hypothesis to the data by collecting various city level measures of social capital.They show that looking across 400 Italian cities, there is a positive relationship betweentheir measures of social capital and whether the city was free in 1176.

Nunn and Wantchekon (2011) consider the historical determinants of trust withinthe African context. The authors examine whether the trans-Atlantic and Indian Oceanslave trades influenced the amount of distrust within society. This is done by combininghousehold survey data with estimates of the number of slaves taken from each ethnicgroup in Africa.The study finds a negative relationship between an individual’s reportedtrust in others (either neighbors, relatives, local governments, co-ethnics, and those from

19 A more recent example is Mokyr’s (2008) argument that an important determinant of the IndustrialRevolution was the development of a social norm he calls“gentlemanly culture”that emphasized honesty,commitment, and cooperation.

382 Nathan Nunn

other ethnicities) and the number of slaves taken from the individual’s ethnic group duringthe slave trades.

The study attempts to distinguish between the two most plausible channels throughwhich the slave trades could have adversely affected trust. One channel is that they alteredthe cultural norms of the ethnic groups exposed to the trade,making them inherently lesstrusting. A second channel is that the slave trades resulted in a long-term deterioration oflegal and political institutions,which causes individuals to be less trusting of others today.

The authors undertake a number of tests to distinguish between these two channels.One test examines individuals’ trust in the local government and attempts to control forthe quality of domestic institutions using the individuals’ perceived quality of the localgovernment, extent of corruption, and whether local councillors listen to their concerns,as well as measures of the quality of public goods provision.

Another test controls for a second measure of slave exports: the average number ofslaves that were taken from the geographic location that each individual is currently livingin. This is different from the first measure, which is the average number of slaves takenfrom an individual’s ethnic group.The second slave export variable is motivated by the factthat when an individual relocates the individual’s internal norms move with them,but theexternal institutional environment is left behind. In other words, institutions, which areexternal to the individual, are much more geographically fixed, relative to cultural beliefswhich are internal to the individual.Therefore,the two variables can be used to distinguishthe extent to which the slave trade affects trust through the culture channel versus throughthe institutions channel. If the slave trade affects trust primarily through internalizednorms and cultural beliefs, which are ethnically based and internal to the individuals,then when looking across individuals, what should matter is whether their ancestorswere heavily enslaved. If the slave trade affects trust primarily through its deterioration ofdomestic institutions, which are external to the individual and geographically immobile,then what should matter is whether the external environment the individual is living inwas heavily affected by the slave trades.

The results of each of the tests indicate that the slave trades adversely affect trustthrough both cultural norms and institutions, but that the magnitude of the culturechannel is always greater than the institutions channel.

Another cultural consequence of the slave trade that has received attention is the prac-tice of polygamy. Because significantly more men than women were enslaved during thetrans-Atlantic slave trade, the ratio of men to women in Africa was significantly affected.It has been hypothesized that this gave rise to the practice of polygamy. CombiningNunn and Wantchekon’s (2011) estimates of ethnicity-level slave exports and informa-tion from household survey data, Dalton and Leung (2011) and Fenske (2012) find apositive relationship between slave exports and the prevalence of polygamy.

Other examples of evidence for the historical origins of current cultural traits includeAlesina et al.’s (2013) study of the relationship between traditional plough use and current


gender roles, as well as Durante’s (2010) analysis of the link between historical weathervariability and current trust. Both have been described earlier in the chapter.

7.5.4 The Interplay Between Culture and InstitutionsGenerally, studies of the historical importance of culture and studies of the historicalimportance of institutions are done in isolation of each other. However, there is evidencethat there are important complementarities and interdependencies between culture andinstitutions. I now turn to a discussion of these.

7.5.4.1 Culture Affecting Formal InstitutionsHistorically, there are many examples of culture impacting the evolution of domesticinstitutions. Arguably, the most obvious are the European migrant communities estab-lished around the globe after the Age of Exploration. At a macro-level, this has beenillustrated by Acemoglu et al.’s (2001) colonial origins hypothesis. A more micro-levelanalysis of the process (at least for the United States) is provided in David HackettFischer’s (1989) book Albion’s Seed,where he demonstrates that the institutions and socialstructures initially established by European migrants arose from the values and beliefsbrought with them from the Old World. In other words, the institutions first establishedwere endogenous to the cultural beliefs of the early migrants.

Fisher documents four waves of early migration to North America—the Puritans(1629–1641), the Anglican Cavaliers (1642–1675), the Quakers (1675–1725), and theScotch-Irish (1717–1775)—and shows how differences in the values of each immigrantwave generated differences in the institutions that were established.The Puritans, believ-ing in the importance of universal education and in a well-functioning society,establisheduniversal education, significant taxes, sizable governments, and heavy-handed justice.TheVirginia Cavaliers,who believed that inequality was natural and were primarily concernedwith maintaining existing forms of hierarchy, implemented limited education, lower taxes,less government spending, and an informal system of justice based on hierarchical vio-lence. The institutions established by the Quakers in the DelawareValley reflected theirbelief in the central importance of personal freedoms.All citizens were granted equal legalrights, there was limited government involvement in personal and religious affairs, andtaxes were limited.The institutions implemented by the Scotch-Irish were an outgrowthof their belief in freedom from the constraints imposed by government.This resulted in alimited formal justice system (and a reliance on ad hoc vigilante justice), limited politicalinstitutions, light taxes, and strong rights to armed resistance from authority.

European mass migration provides one episode that clearly illustrates the endogene-ity of institutions to cultures. Other studies also provide similar evidence from othercontexts. For example, Zerbe and Anderson (2001) document that the initial propertyrights institutions established during the 1848 California Gold Rush reflected the val-ues and beliefs that miners brought with them westward. The beliefs—which included

384 Nathan Nunn

individualism, respect for property, and the view that rewards should be commensuratewith effort—first developed into collectively practiced norms of behavior (i.e. informalinstitutions) before being formalized as written laws.

As well, the work by Greif (1994) on the cultural differences between the Maghribiand Genoese medieval traders also illustrates the role of culture in shaping the forma-tion of formal institutions. The Genoese developed institutions that arose from theirindividualist cultural beliefs, including a formal legal system as well as other formal orga-nizations that helped to facilitate exchange. By contrast, the institutional structures of theMaghribis grew out of their collectivist cultural beliefs. Because the Maghribis continuedto rely on informal enforcement mechanisms, organizations remained limited in size andscope.

7.5.4.2 Institutions Affecting CultureThere is also the possibility of feedback effects, with formal institutions affecting theevolution of cultural traits. A number of recent studies have found evidence for this. Forexample, Guido Tabellini (2010) explains variation across regions of Europe in levels oftrust, respect, and confidence in the returns to individual effort. He identifies a strongpositive relationship between the prevalence of these cultural traits and measures of theaverage quality of domestic institutions between 1600 and 1850.The estimates show thatEuropean regions that had less well-developed institutions in the past have less trust inothers, less respect for others, and believe less in the value of individual effort today.

Evidence for the impact of institutions on culture also comes from a number ofstudies that use a regression discontinuity strategy, focusing on particularly importanthistorical borders that today lie within the same country. Becker et al. (2011) examineEastern European villages lying within the same country today, but on either side ofthe historical Habsburg border. They show that villages that were formerly part of theHabsburg Empire, with its greatly respected and well-functioning bureaucracy, todayhave greater trust in their local government. Grosjean (2011b) examines location pairswithin Eastern Europe and shows that the longer a pair was under the same Empirehistorically, the more similar the reported social trust of the locations’ citizens today.Peisakhin (2010) surveys 1,675 individuals living in 227 villages located within 25 kmof the Habsburg-Russian border that divided Ukraine between 1772 and 1918. Relyingon information on cultural traits based on answers to survey questions, Peisakhin (2010)documents a wide range of statistically significant cultural differences between the twogroups.

7.5.4.3 Coevolution of Culture and InstitutionsTabellini (2008) provides a formal model of the interplay between culture and institutionsin an environment in which both are endogenous and co-evolve. In the model, thereare two potential cultural traits with one valuing cooperation (or believing cooperation


is the right thing to do) more than the other. Vertical transmission of these values ismodeled explicitly with parents exerting costly effort to instill values of cooperation.One of the primary innovations of the paper is to also model the endogenous formationof institutions (which enforce cooperation) through majority voting. Tabellini showsthat the co-evolution of culture and institutions generates strategic complementarity andmultiple equilibria. A culture that values cooperation prefers institutions that stronglyenforce cooperation, which in turn increases the returns to cooperation, reinforcing thiscultural trait. Conversely, a culture that does not value cooperation prefers institutionsthat weakly enforce cooperation, which in turn decreases the returns to cooperation,reinforcing a culture that does not value cooperation.

Recent papers that have empirically studied contemporary institutions and cultureprovide evidence of interactions between culture and institutions. Aghion et al. (2011)examine contemporary labor markets and identify a negative cross-country relationshipbetween the existence of cooperative labor relations and the severity of minimum wageregulation by the state. Similarly,Aghion et al. (2010) identify a negative cross-countryrelationship between general trust and government regulation.

Both studies then develop models of the interplay between institutions/policies andculture/beliefs. In both,greater government regulation crowds out beneficial behavior ofcitizens. In Aghion et al. (2011), higher minimum wage regulation reduces the benefitsto workers of trying to cooperate with firms. Therefore, more stringent minimum wageregulations crowd out cooperation between firms and workers. In turn, less cooperativefirm-worker relationships increase the demand for minimum wage regulation.Thus, thisinterdependence explains the observed negative relationship between minimum wageand cooperative labor relations.

In Aghion et al. (2010), a low level of civic mindedness in the economy results in agreater need for regulation to protect citizens from the negative externalities imposedby those that are not civic-minded. The high level of regulation in the economy alsoreinforces the low level of civic mindedness, as it is these individuals that are comfortablepaying and demanding bribes. The result is that greater trust is observed in economieswith lower levels of government regulation.

What the three studies described here have in common is their analysis of the two-wayrelationship between culture and institutions. Given this interdependence, both institu-tions and culture co-evolve, and this can generate multiple stable equilibria with differentsets of institutions and cultural norms that are self-enforcing.

7.5.5 GeneticsIt is possible that historical events that affect the distribution of individuals in differentlocations—i.e. through genocide,forced expulsions,or voluntary migrations—could havelong-term impacts through a genetic channel. Given that genetic traits tend to be fairly

386 Nathan Nunn

persistent over time, if they have an impact on economic outcomes, then it is theoreticallypossible that events that impact the genetic distribution of the population may have long-term economic impacts.

A number of recent studies provide some evidence that genetics can impact humanbehavior. For example, Cesarini et al. (2008) exploit variation in genetic differencesbetween monozygotic and dizygotic twins.The authors compare differences in the actionstaken in a standard trust game between monozygotic and dizygotic twins from Swedenand the United States. By assuming that similarity of twin behavior can be decomposedinto a common environment, common genes, and other individual-specific variables, andthat monozygotic twins share the same environment and same genes, and dizygotic twinsshare the same environment but have half the alleles of genes, they are able to estimate theextent to which behavior is genetically determined. They find that monozygotic twinsconsistently exhibit more similar behavior than dizygotic twins, and therefore, basedon their assumptions, they conclude that an important part of behavior is geneticallydetermined.The same basic procedure is repeated in Cesarini et al. (2009),but examiningbehavior in a dictator game and measures of individual risk aversion.

At the macro-level, a number of studies have documented relationships betweengenetic measures and economic outcomes. Spolaore and Wacziarg (2010) show thatgreater genetic relatedness has a positive impact on the probability that two populationsgo to war with one another. Spolaore and Wacziarg (2009) show that, across country-pairs, bilateral genetic distance is positively associated with current income differences.In other words, genetically similar countries are economically more similar.

Ashraf and Galor (2012) provide evidence that genetic diversity within a countryis non-monotonically related to per capita income. There is an inverted-U relationshipbetween the two, with income reaching a maximum at an intermediate level of diversity.Too much diversity and too little diversity are both associated with low per capita income.

Cook (2013a) also examines genetic diversity, but unlike Ashraf and Galor (2012),he considers a specific group of genes associated with resistance and susceptibility todisease, namely the major histocompatibility complex. Within humans this is a clusterof 239 genes on the 6th chromosome. Cook (2013a) measures the variation in allelefrequency within this system and shows that, across countries, his measure of geneticvariation is positively correlated with Olsson and Hibbs’s (2005) measure of the numberof domesticable animals and Putterman’s (2008) measure of the time since the adoption ofagriculture. It is also positively correlated with health,measured in the 1960s. Interestingly,by 1990 the health relationship no longer exists.

In a subsequent study,Cook (2013c) considers another channel through which genet-ics could have long-term impacts on economic development. This is through lactasepersistence (i.e. the ability to digest milk after childhood). Cook hypothesizes that, his-torically, societies with the gene variant that resulted in lactase persistence had accessto an additional source of calories, vitamins, and nutrients, which resulted in increased


population densities.The author shows that looking across countries,a greater proportionof the population with lactase persistence is associated with greater population density,measured in 1500.

7.6. UNRESOLVEDQUESTIONS ANDDIRECTIONS FOR FUTURERESEARCH

7.6.1 Persistence or Reversals?A number of studies provide evidence of the persistence in economic development overlong periods of time. Societies that were more economically, technologically, or institu-tionally developed in the past are also the most developed today. For example, Cominet al. (2010) document a positive relationship between historical technology levels (asfar back as 1000 bc) and current income per capita across different parts of the world.Along similar lines, Bockstette et al. (2002) empirically document a positive relationshipbetween state antiquity and current economic performance today. Societies that weremore politically developed in the past, are more economically developed today. At amore micro-level and over a shorter timespan, Huillery’s (2011) analysis of French WestAfrica shows persistence of prosperity between the pre- and post-colonial period.

These findings of persistence stand in contrast to the“reversal of fortunes”documentedin Acemoglu et al. (2002): among a sample of former colonies, those locations that werethe most prosperous in 1500 are the most underdeveloped today. This reversal has alsobeen confirmed in alternative studies. For example, Nunn (2008a) shows that amongAfrican countries, those that were the most developed prior to the slave trades (measuredby population density in 1400) had the largest number of slaves taken and have the lowestincomes today.20

These two sets of findings appear to stand in contrast with one another, one showingpersistence over long periods of time and the other showing a complete reversal. Whichis correct? It turns out that both are, and an important part of the difference arises dueto differences in the samples being examined. The persistence studies tend to examineall countries globally, while the reversal studies have samples that only include formercolonies.

To illustrate this, consider the bivariate relationship between the natural log of percapita income in 1500 and the natural log of real per capita GDP in 2000.This relationshipis reported in columns (1)–(3) of Table 7.1.The sample comprises 85 former colonies and65 non-colonies examined in Comin et al. (2010). Column (1) reports the relationshipamong former colonies.This is analogous to the regressions estimated by Acemoglu et al.(2002). As shown, consistent with their findings, there is a negative relationship between

20 Also related is the question of whether Africa has always been behind the rest of the world. While theconventional wisdom is that Africa has generally been the most underdeveloped continent of the world,there is evidence that this view is misplaced (Ehret, forthcoming).

388 Nathan Nunn

population density in 1500 and per capita income today. There has been a reversal.Column (2) examines this same relationship among the rest of the sample, which are thecountries that were never colonized. Here one observes a very different relationship.Thetwo variables are positively correlated.Among this group there is persistence. Column (3)reports the relationship between the full sample, and shows that on average there ispersistence.The coefficient is positive and significant at the 10 percent level.This estimateis analogous to the findings of persistence by Comin et al. (2010),Bockstette et al. (2002),and others.

Acemoglu et al. (2002) argue that among former colonies, the reversal occurredbecause initial prosperity impacted the institutions that were developed by Europeans.Where initial incomes were low, population was sparse, and Europeans settled, establish-ing protection of property rights and other growth-promoting institutions.Where initialincomes were high, Europeans undertook an extractive strategy. In some cases, theyco-opted existing forced labor traditions, and in others, they promoted enslavement andthe sale of indigenous populations. As a result, locations that were initially poor in 1500today are more developed than those that were initially richer.

A similar but alternative explanation for the reversal, and one that has been stressed inthe recent paper by Easterly and Levine (2012), is that the less populated places witnessan in-migration of people from more prosperous countries, with higher levels of humancapital, culture more conducive to economic growth, and/or other vertically transmittedtraits.Therefore these locations are richer today.This alternative explanation suggests thatthe reversal simply reflects migration and the persistence of prosperity at the society level.

This alternative explanation can be examined using an ancestry-based measure ofpopulation density in 1500 and per capita GDP in 2000. The ancestry-based initialpopulation measure is constructed using Putterman and Weil’s (2010) World MigrationMatrix.While the geography-based measure used in columns (1)–(3) is the average income(proxied by population density) of people living on the country’s land in 1500, theancestry-based measure is the average income in 1500 (proxied by population density)of the ancestors of those living in the country today.

The estimates from column (4) show that, all else equal, among former colonies,being descended from ancestors with a high prosperity is positively associated with percapita income today. Therefore, it is plausible that colonial migration of individuals fromprosperous societies explains the reversal. Interestingly, the persistence of income alonglineages (and not locations) is similarly strong among non-colonies (for which there isless migration) as among former colonies (with much greater migration).21

A simple way to examine whether the reversal documented in Acemoglu et al. (2002)is explained by migration combined with the persistence of prosperity across generations

21 Better understanding the specific transmission mechanisms underlying this persistence is the subject ofongoing research and debate. For an excellent summary of this literature see Spolaore and Wacziarg(forthcoming).

HistoricalD

evelopment

389

Table 7.1 Persistence and reversals

(1) (2) (3) (4) (5) (6) (7)Dependent variable: ln per capita GDP in 2000

Colonies Non-colonies Both Colonies Non-colonies Colonies Non-colonies

ln population density in 1500:Geography-based −0.228∗∗∗ 0.276∗∗∗ 0.115∗ −0.316∗∗∗ 0.003

(0.070) (0.090) (0.061) (0.058) (0.319)Ancestry-based 0.475∗∗∗ 0.319∗∗∗ 0.581∗∗∗ 0.316

(0.098) (0.100) (0.086) (0.355)

Observations 85 65 150 85 65 85 65R-squared 0.114 0.129 0.023 0.222 0.140 0.430 0.140

Notes: The table reports coefficients from OLS estimates,with standard errors in parentheses.The dependent variable is the natural log of real per capita GDP in 2000.The independent variables are the natural log of a country’s population density in 1500,measured as the historical average of the land of the country (geography-based)or the historical average of the ancestors of the population of the country (ancestry-based). The correlation between the two measures is 0.23 for colonies and 0.96for non-colonies.∗ Indicate significance at the 10% level.∗∗ Indicate significance at the 5% level.∗∗∗ Indicate significance at the 1% level.

390 Nathan Nunn

is to examine the coefficient of the geography-based measure of population density in1500, while controlling for the ancestry-based measure. This is done in column (6). Themagnitude of the coefficient for the geography-based measure of 1500 prosperity doesnot diminish, and actually increases. The ancestry-based 1500 prosperity measure enterswith a large positive and significant coefficient. This suggests the coexistence of twochannels. One is the migration of populations from more prosperous societies and theother being the reversal of fortune discussed in Acemoglu et al. (2002). This finding ofstronger persistence by ancestry than by location is not new and is an important pointmade in Putterman and Weil (2010), Comin et al. (2010), and Chanda et al. (2013).

Column (7) examines the same correlations as column (6), but among the sampleof non-colonies. Because there is little migration among this group, the two populationdensity measures are highly correlated (the correlation coefficient is 0.96). Due to thismulticollinearity, both variables are insignificant. However, the estimated coefficient forthe ancestry-based variable provides evidence of persistence across generations that issimilar in magnitude but smaller than the estimate for the colonies sample. As expected,the coefficient for the geography-based variable shows no evidence of a reversal-of-fortunes mechanism among non-colonies.

Overall, the correlations reported in Table 7.1 are suggestive of the following facts.First, within former colonies, there has been a reversal of fortunes (looking at geographiclocations as the unit of observation). Second,no such reversal exists among non-colonies.Third, there is no reversal once one uses societies (and their descendants) as the unit ofobservation. Instead one observes extreme persistence, both among former colonies andnon-colonies, a fact that has been empirically noted by Putterman and Weil (2010) anddiscussed in Spolaore and Wacziarg (forthcoming). Fourth, the Acemoglu et al. (2002)reversal exists even after accounting for the migration of populations from more prosper-ous to less prosperous regions during the colonial period. This does not appear to fullyexplain the reversal.

Therefore,the existence of reversals and persistence in the data seem to be reconcilable.However, the most recent research along these lines shows a reversal that is not explainedby the logic above. Olsson and Paik (2012) document a reversal within Europe fromthe Neolithic until now. They show that the parts of Europe that adopted agricultureearlier (and were arguably more economically developed during the Neolithic period)are less developed today. Although the authors provide an explanation, the exact reasonfor this reversal is far from clear.They also find evidence of a reversal within sub-SaharanAfrica and East Asia. The reason behind the reversals in these regions is also unclear.Most interestingly, they show that if one looks at a global sample, there is persistence: theparts of the world that adopted agriculture earlier are more developed today. In otherwords, looking within-regions there are reversals, but looking across regions (and acrosscountries generally) there is persistence.


7.6.2 When Doesn’t History Persist?To date, the primary focus of the literature has been in empirically documenting thepersistence of historical shocks, typically arising due to lasting impacts through eitherdomestic institutions or cultural traits. Little or no attention has been placed on examiningwhen historical events do not have lasting impacts.This emphasis is logical given the needto first establish that history can matter, which has led to a natural focus on events thathave had persistent impacts.

However, there are a few studies that provide some preliminary evidence for whenhistory persists and when it does not. For example,Voigtlaender andVoth (2012) docu-ment the persistence of anti-Semitic values and beliefs in Germany between the 14th and20th centuries. Their analysis examines variation across German villages and documentsa remarkable relationship between the prevalence of pogroms during the Black Death(1348–1350) and a number of measures of anti-Semitic sentiment in the early 20thcentury. The authors then turn to an analysis of the environments in which persistencewas more or less strong. One of their most interesting findings is that the persistenceof this cultural trait is much weaker among Hanseatic cities, which were self-governedGerman cities heavily involved in lucrative long-distance trade. This finding may bedue to higher rates of migration or to more dynamism arising from greater economicopportunity and growth. Voigtlaender and Voth (2012) also find that (consistent withboth mechanisms) there is less persistence among cities with faster population growth,and (consistent with the second mechanism) there is less persistence among cities thatwere more industrialized in 1933.

Grosjean’s (2011a) study of Nisbett and Cohen’s (1996)“culture of honor”hypothesisshows a persistent impact of the Scotch-Irish culture of honor, but only within theSouthern states of the US.The obvious explanation is that a cultural heuristic of aggressionwas relatively beneficial in the south,which was more lawless and with less well-developedproperty rights institutions. However, in the north, with a more established rule of lawand better developed property rights protection, norms of aggression and violence wereless beneficial, and therefore did not persist. In other words, external characteristics—inthis case domestic institutions—by affecting the relative costs and benefits of differentcultural norms, influence their persistence.

Another environment in which this can be seen is in Africa in the context of theslave trade. A natural hypothesis is that the detrimental impacts of the slave trades ontrust will be more persistent in countries with a poorly functioning legal system. It is inthese environments, where individuals are not legally constrained to act in a trustworthymanner, that norms of mistrust, initially developed by the slave trade, may continue to berelatively beneficial and to persist.

This can be tested directly by re-estimating Equation (7.1) from Nunn andWantchekon(2011), but allowing for the impact of past slave exports on trust today to depend on thequality of country-level domestic institutions, measured at the time of the survey (2005)

392 Nathan Nunn

using the Governance Matters “rule of law” variable. The original index ranges from−2.5 to +2.5, but I normalize the variable to lie between zero and one.22 The aug-mented equation is:

trusti,e,d,c = αc + β1 slave exportse + β2 slave exportse × rule of lawc

+X′i,e,d,cΓ + X′

d,cΩ + X′eΦ + εi,e,d,c , (7.1)

where i indexes individuals, e ethnic groups, d districts, and c countries; trusti,e,d,c denotesone of five individual-level measures of trust that range from 0 to 3; slave exportse is ameasure of the number of slaves taken from ethnic group e during the Indian Oceanand trans-Atlantic slave trades23; rule of lawc is the 0-to-1 measure of a country’s ruleof law in 2005; αc denotes country fixed effects; and X′

i,e,d,c , X′d , and X′

e denote vectorsof individual-, district-, and ethnicity-level control variables. See Nunn andWantchekon(2011) for a fuller description of the variables in Equation (7.1).

Estimates of Equation (7.1) are reported in Table 7.2. The table reports estimates ofβ1 and β2. The bottom panel reports estimates of the impact of the slave trade on trustfor the country with the lowest measure of rule of law (0.17) and for the country withthe highest rule of law measure (0.63).24 As shown, the estimated coefficient for theinteraction term β2 is positive in all specifications (although the precision of the estimatevaries). This indicates a weaker negative impact of the slave trades on trust in countrieswith better domestic institutions. Further, for all trust measures, the estimated impact ofthe slave trades on trust is positive and significant for the lowest rule of law country butnot statistically different from zero for the highest rule of law country. This is consistentwith the adverse impacts of the slave trade being less persistent in countries with a betterrule of law. In these countries,well-functioning institutions enforce trustworthy behaviorof its citizens and therefore there is less persistence of the mistrust engendered by theslave trades.

An important shortcoming of this exercise arises due to the endogeneity of thecountry-level rule of law measure. In particular, it is likely endogenous to the slavetrade. Ideally, estimates of this nature would rely on exogenous variation in the variableused to test for heterogeneity. However, an important point to bear in mind is that theestimates reported in Table 7.2 and Nunn and Wantchekon (2011) are estimated usingwithin-country variation only. Any impacts that the slave trade had on country-levelcharacteristics are controlled for directly in the regression because of the presence ofcountry fixed effects. In other words, although the rule of law measure is an endogenous

22 This is done by adding 2.5 to the measure and dividing by 5.23 The measure is the natural log of one plus total slave exports normalized by land area.24 Zimbabwe is the country with the lowest rule of law measure in theAfrobarometer sample,and Botswana

is the country with the highest.

HistoricalD

evelopment

393

Table 7.2 Testing for heterogenous impacts of the slave trade on trust in Nunn and Wantchekon (2011)

Trust of relatives Trust of neighbors Trust of local council Intra-group trust Inter-group trust(1) (2) (3) (4) (5)

ln (1 + exports/area) −0.172 −0.341∗∗∗ −0.170∗∗∗ −0.461∗∗∗ −0.344∗∗∗(0.141) (0.115) (0.064) (0.102) (0.082)

ln (1 + exports/area) × rule of law 0.111 0.512∗ 0.169 0.891∗∗∗ 0.695∗∗∗Index 2005 (0.360) (0.302) (0.173) (0.263) (0.208)

Individual controls Yes Yes Yes Yes YesDistrict controls Yes Yes Yes Yes YesCountry fixed effects Yes Yes Yes Yes Yes

Number of observations 20,062 20,027 19,733 19,952 19,765Number of ethnicity clusters 185 185 185 185 185Number of district clusters 1257 1257 1283 1257 1255

Estimated impact for Afrobarometer −0.153∗ −0.252∗∗∗ −0.141∗∗∗ −0.305∗∗∗ −0.223∗∗∗country with lowest rule of law (0.082) (0.066) (0.036) (0.059) (0.049)

Estimated impact for Afrobarometer −0.102 −0.018 −0.064 0.102 0.095country with highest rule of law (0.098) (0.087) (0.054) (0.073) (0.058)

Notes:The table reports OLS estimates. The unit of observation is an individual. ln (1 + exports/area) is the number of slaves exported normalized by land area,measured at the ethnicity level. Rule of Law Index 2005 is the Governance Matters VI rule of law measure for 2005, normalized to lie between zero and one.Standard errors are adjusted for two-way clustering at the ethnicity and district levels. The individual controls are for age, age squared, a gender indicator variable,5 living conditions fixed effects, 10 education fixed effects, 18 religion fixed effects, 25 occupation fixed effects, and an indicator for whether the respondent livesin an urban location. The district controls include ethnic fractionalization in the district and the share of the district’s population that is the same ethnicity as therespondent. See Nunn and Wantchekon (2011) for further details.∗ Indicate significance at the 10% level.∗∗ Indicate significance at the 5% level.∗∗∗ Indicate significance at the 1% level.

394 Nathan Nunn

variable, its direct (linear) impacts on trust are captured by the country fixed effects inthe regression.

Looking at the differences in the estimates of β1 and β2 across the five trust measures, itis clear that the heterogeneous impacts of the slave trades are weaker for trust of relativesand trust of the local government. This is true whether one considers the magnitudeand significance of β2 or of the high and low estimates reported in the bottom panelof the table. Interestingly, disputes between relatives and disputes between citizens andthe local government are less likely to be resolved through the legal system than disputesbetween neighbors, co-ethnics, or citizens from different ethnic groups. Given this, wewould expect that rule of law would be less successful in enforcing good behavior inthese situations, and as a result would be a less important determinant of the persistenceof distrust. The estimates reported in Table 7.2 are consistent with this.

7.7. CONCLUSIONS: LOOKING BACKWHILEMOVING FORWARD

This chapter has provided a broad overview of research examining comparativehistorical economic development. Studies have examined a wide array of historic events,including the Neolithic Revolution, colonial rule, Africa’s slave trades, the IndustrialRevolution, the Protestant Reformation, the French Revolution, and the ColumbianExchange.

Although the studies reviewed in this chapter have done much to identify importantpieces of the larger historical puzzle, many of the pieces are yet to be uncovered. Inaddition, the more difficult task is understanding exactly how all of the pieces fit together.This is a step that has not been taken by the vast majority of the studies in the literaturesummarized here. Nearly all examine a particular event in isolation from other events,except possibly to account for other events as covariates in the empirical analysis. However,once one begins thinking of the realities of history, it is soon apparent that historical eventsimpact other historical events in important and sometimes subtle or complicated ways.Further, there are often complex interactions between events, suggesting that the linearspecifications typically assumed in studies may be inaccurate.

There are many examples of these interdependencies. For example, Europe’s abilityto colonize and rule the African continent depended critically on the discovery of thechincona tree in the Andes and its mass production in Asia by the British.This is becausequinine, the first effective protection against malaria, is derived from the bark of the tree.Similarly, European knowledge of how to effectively process wild rubber obtained fromNative Americans had important consequences for the millions of Africans that weretortured and killed in King Leopold’s Congo.

Another example is the interdependence between the printing press and both theProtestant Reformation (Dittmar, 2011; Rubin, 2011) and the Atlantic trade (Dittmar,


2011).We have also seen that Catholic conflict with the Ottoman Empire helped enablethe spread of the Protestant religion across Europe (Iyigun, 2008).

We have seen that the presence of the tsetse fly in Africa resulted in less intensiveagriculture that did not use animals or the plough (Alsan, 2012). Because the ploughwas not adopted, women participated actively in agriculture, which generated normsof equality, which continue to persist today (Alesina et al. 2013). This an importantexplanation for the high levels of female labor force participation that is observed inAfrica today.

We have seen that Africa’s slave trades resulted in underdeveloped pre-colonial states(Nunn, 2008a), which in turn are associated with less post-colonial public goods provi-sion and lower incomes (Gennaioli and Rainer, 2007; Michalopoulos and Pappaioannou,2013).

Moving forward, the second major task for the literature to tackle is to better under-stand channels of causality. In the past decade, we have made significant progress empir-ically testing whether historical events have lasting impacts. The bulk of this surveyis devoted to reviewing this evidence, which overwhelmingly shows that history doesmatter. What is less clear is exactly why it matters. I have reviewed here the leadingcandidates:multiple equilibria, cultural norms of behavior, and domestic institutions.Theextent to which these mechanisms matter, and in which circumstances, is yet to be fullyunderstood. Further, as discussed, there are also potentially important complementaritiesbetween the channels. For example, beliefs and values tend to become codified in formalinstitutions, which in turn feedback, affecting the evolution of these values. Comple-mentarities between cultural traits and formal institutions are likely an important part ofmany instances of long-term persistence.

Overall, while much progress has been made to this point, the primary accomplish-ment has been in establishing the importance of studying the past for understandingcurrent growth and development. The economic literature is increasingly coming tounderstand that where we are (and therefore how we best move forward) has a lot to dowith how we got here.

ACKNOWLEDGMENTThe chapter has benefitted from helpful discussions with Emmanuel Akyeampong, Robert Bates, SamuelBowles, Claudia Goldin, Joseph Henrich, and James Robinson. I thank Eva Ng for research assistance.

REFERENCES

Acemoglu, Daron, 1995. Reward structure and the allocation of talent. European Economic Review 39,17–33.

Acemoglu, Daron, Johnson, Simon, 2004. Unbundling Institutions. Journal of Political Economy 113, 949–995.

Acemoglu, Daron, Johnson, Simon, Robinson, James A., 2001. The colonial origins of comparative devel-opment: an empirical investigation. American Economic Review 91, 1369–1401.

http://refhub.elsevier.com/B978-0-444-53538-2.00007-1/h0005



396 Nathan Nunn

Acemoglu, Daron, Johnson, Simon, Robinson, James A., 2002. Reversal of fortune: geography and institu-tions in the making of the modern world income distribution. Quarterly Journal of Economics 117,1231–1294.

Acemoglu,Daron, Johnson, Simon,Robinson, James A., 2005.The rise of Europe: atlantic trade, institutionalchange and economic growth. American Economic Review 95, 546–579.

Acemoglu, Daron, Bautista, María Angélica, Querubin, Pablo, Robinson, James A., 2008. Economic andpolitical inequality in development: the case of Cundinamarca, Colombia. In: Helpman, Elhanan (Ed.),Institutions and Economic Performance, Harvard University Press, Cambridge, MA, pp. 181–245.

Acemoglu,Daron,Cantoni,Davide, Johnson,Simon,Robinson, JamesA.,2011a.The consequences of radicalreform: the french revolution. American Economic Review 101 (7), 3286–3307.

Acemoglu, Daron, Hassan,Tarek A., Robinson, James A., 2011b. Social structure and development: a legacyof the Holocaust in Russia. Quarterly Journal of Economics 126 (2), 895–946.

Acemoglu, Daron, Johnson, Simon, Robinson, James A., 2012. The colonial origins of comparative devel-opment: an empirical investigation: reply. American Economic Review 102 (6), 3077–3110.

Ager, Philipp, Ciccone, Antonio, 2012. Rainfall Risk and Religious Membership in the Late Nineteenth-Century US. Universitat Pompeu Fabra, Mimeo.

Aghion, Philippe, Algan,Yann, Cahuc, Pierre, Shleifer, Andrei, 2010. Regulation and Distrust. QuarterlyJournal of Economics 125 (3), 1015–1049.

Aghion,Philippe,Algan,Yann,Cahuc,Pierre,February 2011. Civil society and the state: the interplay betweencooperation and minimum wage regulation. Journal of the European Economic Association 9 (1), 3–42.

Aghion, Philippe, Persson,Torsten, Rouzet, Dorothee, 2012. Education and Military Rivalry. Harvard Uni-versity, Mimeo.

Akçomak, I. Semih,Webbink, Dinand,Weel, Bas ter, 2012. Why did the Netherlands develop so early? TheLegacy of the Brethren of the Common Life, Mimeo.

Albouy, DavidY., 2012. The colonial origins of comparative development: an empirical investigation: com-ment. American Economic Review 102 (6), 3059–3076.

Alesina, Alberto, Giuliano, Paola, Nunn, Nathan, 2013. On the origins of gender roles: women and theplough. Quarterly Journal of Economics 128 (2), 155–194.

Alsan, Marcella, 2012. The Effect of the TseTse Fly on African Development. Harvard University, Mimeo.Andersen,Thomas Barnebeck, Bentzen, Jeanet, Dalgaard, Carl-Johan, 2011. Religious Orders and Growth

Through Cultural Change in Pre-Industrial England. University of Copenhagen, Mimeo.Andersen, Thomas Barnebeck, Jensen, Peter Sandholt, Skovsgaard, Christian Stejner, 2013. The Heavy

Plough and the European Agricultural Revolution in the Middle Ages: Evidence from a HistoricalExperiment. University of Southern Denmark, Mimeo.

Ashraf, Quamrul, Galor, Oded, 2012. The out of Africa hypothesis. Human genetic diversity, and com-parative economic development. American Economic Review 103 (1), 1–46.

Ashraf, Quamrul, Michalopoulos, Stelios, 2011. The Climatic Origins of the Neolithic Revolution:Theoryand Evidence. Brown University, Mimeo.

Austin, Gareth, November 2008. The “Reversal of Fortune” thesis and the compression of history: perspec-tives from African and comparative economic history. Journal of International Development 20 (8),996–1027.

Bai,Ying, Kung, James Kai-sing, 2011a. Climate shocks and sino-nomadic conflict. Review of Economicsand Statistics 93 (3), 970–981.

Bai,Ying,Kung, James Kai-sing, 2011b. Diffusing Knowledge while Spreading God’s Message:Protestantismand Economic Prosperity in China, 1840–1920. Hong Kong University of Science and Technology,Mimeo.

Banerjee,Abhijit, Iyer,Lakshmi,2005. History, institutions and economic performance: the legacy of colonialland tenure systems in India. American Economic Review 95 (4), 1190–1213.

Banerjee, Abhijit, Iyer, Lakshmi, 2008. Colonial Land Tenure, Electoral Competition and Public Goods inIndia. Harvard Business School Working Paper 08–062.

Banerjee, Abhijit, Iyer, Lakshmi, Somanathan, Rohini, 2005. History, social divisions and public goods inrural India. Journal of the European Economic Association Papers and Proceedings 3 (2–3), 639–647.

















Baten, Joerg, van Zanden, Jan Luiten, 2008. Book production and the onset of modern economic growth.Journal of Economic Growth 13, 217–235.

Bates, Robert H., forthcoming. The imperial peace in colonial Africa and Africa’s underdevelopment. In:Akyeampong, Emmanuel, Bates, Robert H., Nunn, Nathan, Robinson, James A. (Eds.),Africa’s Devel-opment in Historical Perspective, Cambridge University Press, Cambridge.

Becker, Sascha O.,Woessmann, Ludger, 2008. Luther and the girls: religious denomination and the femaleeducation gap in nineteenth-century Prussia. Scandinavian Journal of Economics 110 (4), 777–805.

Becker, Sascha O., Woessmann, Ludger, 2009. Was weber wrong? a human capital theory of protestanteconomic history. Quarterly Journal of Economics 124 (2), 531–596.

Becker,Sascha O.,Boeckh,Katrin,Hainz,Christa,Woessmann,Ludger,2011.The Empire is Dead,Long Livethe Empire! Long-Run Persistence of Trust and Corruption in the Bureaucracy. Warwick University,Mimeo.

Bentzen, Jeanet Sinding, Kaarsen, Nicolai, Wingender, Asger Moll, 2012. Irrigation and Autocracy. Uni-versity of Copenhagen, Mimeo.

Bleakely, Hoyt, Lin, Jeffrey, 2012. Portage and path dependence. Quarterly Journal of Economics 127, 587–644.

Bockstette,Valeri, Chanda,Areendam, Putterman, Louis, 2002. States and markets: the advantage of an earlystart. Journal of Economic Growth 7, 347–369.

Boserup, Ester, 1970. Woman’s Role in Economic Development. Allen and Unwin, London.Botticini,Maristella,Eckstein,Zvi,2005. Jewish occupational selection:education,restrictions,or minorities?

Journal of Economic History 65 (4), 922–948.Botticini, Maristella, Eckstein, Zvi, 2007. From farmers to merchants, voluntary conversions and diaspora:

a human capital interpretation of Jewish history. Journal of the European Economic Association 5 (5),885–926.

Boyd,Robert,Richerson,Peter J., 1985. Culture and the Evolutionary Process. University of Chicago Press,London.

Bruhn, Miriam, Gallego, Francisco A., 2012. Good, bad, and ugly colonial activities: do they matter foreconomic development? Review of Economics and Statistics 94 (2), 433–461.

Cantoni, Davide, May 2012. Adopting a new religion: the case of Protestantism in 16th century Germany.Economic Journal 122 (560), 502–531.

Cavalli-Sforza, L.L., Feldman, M.W., 1981. Cultural Transmission and Evolution:A Quantitative Approach.Princeton University Press, Princeton.

Cesarini, David, Dawes, Christopher T., Fowler, James H., Johannesson, Magnus, Lichtenstein, Paul,Wallace,Bjorn, March 2008. Heritability of cooperative behavior in the trust game. Proceedings of the NationalAcademy of Sciences 105 (10), 3721–3726.

Cesarini, David, Dawes, Christopher T., Fowler, James H., Johannesson, Magnus, Lichtenstein, Paul,Wallace,Bjorn, May 2009. Genetic variation in preferences for giving and risk taking. Quarterly Journal ofEconomics 124 (2), 809–842.

Chanda,Areendam, Cook, C. Justin, Putterman, Louis, 2013. Persistence of Fortune:Accounting for Popu-lation Movements,There was no Post-Columbian Reversal. Brown University, Mimeo.

Chaney, Eric, 2008. Ethnic Cleansing and the Long-Term Persistence of Extractive Institutions: Evidencefrom the Expulsion of the Moriscos. Harvard University, Mimeo.

Chaney, Eric, Hornbeck, Richard, 2013. Economic Growth in the Malthusian Era: Evidence from the 1609Spanish Expulsion of the Moriscos. Harvard University, Mimeo.

Chaney, Eric, 2013. Revolt on the Nile: economic shocks, religion and political power. Econometrica.Chen, Shuo, Kung, James Kai-sing, 2012. The Malthusian Quagmire: Maize and Population Growth in

China, 1550–1910. Hong Kong University of Science and Technology, Mimeo.Clark, Gregory, 2005. The condition of the working class in England, 1209–2004. Journal of Political

Economy 113 (6), 707–736.Cohen,Dov,Nisbett,Richard E.,Bowdle,Brian F.,Schwarz,Norbert,1996. Insult,agression,and the southern

culture of honor: an “Experimental Ethnography”. Journal of Personality and Social Psychology 70 (5),945–960.


















398 Nathan Nunn

Comin, Diego, Easterly,William, Gong, Erick, 2010. Was the wealth of nations determined in 1000 B.C.?American Economic Journal: Macroeconomics 2 (3), 65–97.

Cook, C. Justin, 2013a. The Long Run Health Effects of the Neolithic Revolution:The Natural Selectionof Infectious Disease Resistance.Yale University, Mimeo.

Cook, C. Justin, 2013b. Potatoes, Milk, and the Old World Population Boom.Yale University, Mimeo.Cook, C. Justin, 2013c. The Role of Lactase Persistence in Precolonial Development. Yale University,

Mimeo.Dalton, J., Leung, T., 2011. Why is Polygamy more Prevalent in Western Africa? An African Slave Trade

Perspective. Wake Forest University, Mimeo.David, Paul A., 1985. Clio and the economics of QWERTY. American Economic Review Papers and

Proceedings 75 (2), 332–337.Davis, Donald R.,Weinstein, David E., 2002. Bones, bombs, and breakpoints: the geography of economic

activity. American Economic Review 92 (5), 1269–1289.Davis, Donald R.,Weinstein, David E., 2008. A search for multiple equilibria in urban industrial structure.

Journal of Regional Science 48 (1), 29–65.Deconinck, Koen,Verpoorten, Marijke, 2013. Narrow and scientific replication of “The Slave Trade and

the Origins of Mistrust in Africa”. Journal of Applied Econometrics.Dell, Melissa, 2010. The persistent effects of Peru’s mining mita. Econometrica 78 (6), 1863–1903.Dell, Melissa, 2012. Insurgency and Long-Run Development: Lessons from the Mexican Revolution. Har-

vard University, Mimeo.Diamond, Jared, 1997. Guns, Germs, and Steel. W.W. Norton & Company, NewYork.Diamond, Jared, 2002. Evolution, consequences and future of plant and animal domestication. Nature 418,

700–707.Dippel, Christian, 2011. Coexistence, Forced Coexistence and Economic Development: Evidence from

Native American Reservations. University of California Los Angeles, Mimeo.Dittmar, Jeremiah E., 2011. Information technology and economic change: the impact of the printing press.

Quarterly Journal of Economics 126 (3), 1133–1172.Dittmar, Jeremiah E., 2012. The Welfare Impact of a New Good:The Printed Book. American University,

Mimeo.Durante, Ruben, 2010. Risk, Cooperation and the Economic Origins of Social Trust:An Empirical Inves-

tigation. Science Po, Mimeo.Easterly, William, Levine, Ross, 2003. Tropics, germs and crops: how endowments influence economic

development. Journal of Monetary Economics 50, 3–39.Easterly,William, Levine, Ross, 2012. The European Origins of Economic Development. NBER Working

Paper 18162.Ehret, Christopher, forthcoming. Africa in world history before c. 1440. In: Akyeampong, Emmanuel,

Bates, Robert H., Nunn, Nathan, Robinson, James A. (Eds.), Africa’s Development in Historical Per-spective, Cambridge University Press, Cambridge, (Chapter 2).

Engerman, Stanley L., Sokoloff, Kenneth L., 1997. Factor endowments, institutions, and differential paths ofgrowth among new world economies: a view from economic historians of the United States. In:Harber,Stephen (Ed.), How Latin America Fell Behind, Stanford University Press, Stanford, pp. 260–304.

Engerman, Stanley L., Sokoloff, Kenneth L., 2002. Factor Endowments, Inequality, and Paths of Develop-ment Among New World Economies,Working Paper 9259, National Bureau of Economic Research.

Engerman, Stanley L., Sokoloff, Kenneth L., 2005. The evolution of suffrage institutions in the Americas.Journal of Economic History 65, 891–921.

Fairbank, John King,1953.Trade and Diplomacy on the China Coast, 1842–1854. Harvard University Press,Cambridge.

Feir, Donna, 2013. The Long Term Effects of Indian Residential Schools on Human and Cultural Capital.University of British Columbia, Mimeo.

Fenske, James, 2011. Ecology,Trade and States in Pre-Colonial Africa. Oxford University, Mimeo.Fenske, James, 2012. African Polygamy: Past and Present. Oxford University, Mimeo.Fenske, James, Kala, Namrata, 2013. Climate, Ecosystem Resilience and the Slave Trade. CEPR Discussion

Paper 9449.















Feyrer, James D., Sacerdote, Bruce, 2009. Colonialism and modern income: islands as natural experiments.Review of Economics and Statistics 91 (2), 245–262.

Fischer, David Hackett, 1989. Albion’s Seed: Four British Folkways in America. Oxford University Press,NewYork.

Gennaioli, Nicola, Rainer, Ilia, 2007. The modern impact of precolonial centralization in Africa. Journal ofEconomic Growth 12 (3), 185–234.

Gimbutas, Marija, 2007. The Goddesses and Gods of Old Europe: Myths and Cult Images. University ofCalifornia Press, Berkeley.

Glaeser, Edward L., Porta, Rafael La, Lopez-De-Silanes, Florencio, Shleifer, Andrei, 2004. Do institutionscause growth? Journal of Economic Growth 9, 271–303.

Greif,Avner, 1993. Contract enforceability and economic institutions in early trade: the Maghribi Traders’coalition. American Economic Review 83 (3), 525–548.

Greif,Avner, 1994. Cultural beliefs and the organization of society: a historical and theoretical reflection oncollectivist and individualist societies. Journal of Political Economy 102 (5), 912–950.

Grennes,Thomas, 2007. The columbian exchange and the reversal of fortune. Cato Journal 27 (1), 91–107.Grosjean, Pauline, 2011a. A History of Violence:The Culture of Honor as a Determinant of Homicide in

the US South. University of New South Wales, Mimeo.Grosjean,Pauline, 2011b.The weight of history on European cultural integration: a gravity approach. Amer-

ican Economic Review Papers and Proceedings 101 (3), 504–508.Guiso, Luigi, Sapienza, Paola, Zingales, Luigi, 2008. Long-Term Persistence, Mimeo.Haber, Stephen, Menaldo,Victor, 2010. Rainfall and Democracy. Stanford University, Mimeo.Hansen, Casper Worm, Jensen, Peter Sandholt, Skovsgaard, Christian, 2012. Gender Roles and Agricultural

History:The Neolithic Inheritance. Aarhus University, Mimeo.Hersch, Jonathan,Voth, Hans-Joachim, 2009. Sweet Diversity: Colonial Goods and the Rise of European

Living Standards after 1492. Universitat Pompeu Fabra, Mimeo.Heywood, Linda, 2009. Slavery and its transformation in the kingdom of Kongo: 1491–1800. Journal of

African History 21, 1–22.Hodder, Ian, January 2005. Women and men at Çatalhöyük. Scientific American 15, 34–41.Huillery, Elise, 2009. History matters: the long-term impact of colonial public investments in French West

Africa. American Economic Journal:Applied Economics 1 (2), 176–215.Huillery, Elise, 2011. The impact of european settlement within French West Africa: did pre-colonial pros-

perous areas fall behind? Journal of African Economies 20 (2), 263–311.Inikori, Joseph E., 2000. Africa and the trans-atlantic slave trade. In: Falola,Toyin (Ed.), Africa Volume I:

African History Before 1885, Carolina Academic Press, North Carolina, pp. 389–412.Inikori, Joseph E., 2002. Africans and the Industrial Revolution in England:A Study in International Trade

and Economic Development. Cambridge University Press, Cambridge.Inikori, Joseph E., 2003. The struggle against the trans-atlantic slave trade. In: Diouf,A. (Ed.), Fighting the

Slave Trade:West African Strategies, Ohio University Press,Athens, Ohio, pp. 170–198.Iyer, Lakshmi, 2010. Direct versus indirect colonial rule in India: long-term consequences. Review of

Economics and Statistics 92 (4), 693–713.Iyigun, Murat, 2008. Luther and Suleyman. Quarterly Journal of Economics 123 (4), 1465–1494.Jancec, Matija, 2012. Do Less Stable Borders Lead to Lower Levels of Political Trust? Empirical Evidence

from Eastern Europe. University of Maryland at College Park, Mimeo.Jha, Saumitra, 2008. Trade, Institutions and Religious Tolerance: Evidence from India. Stanford University,

Mimeo.Jia, Ruixue, forthcoming. The legacies of forced freedom: China’s treaty ports. Review of Economics and

Statistics.Jia, Ruixue, forthcoming.Weather shocks, sweet potatoes and peasant revolts in historical China. Economic

Journal.Lal, R., Reicosky, D.C., Hanson, J.D., 2007. Evolution of the plow over 10,000Years and the rationale for

no-till farming. Soil and Tillage Research 93 (1), 1–12.La Porta, Rafael, Lopez-de-Silanes, Florencio, Shleifer,Andrei,Vishny, Robert, 1997. Legal determinants of

external finance. Journal of Finance 52, 1131–1150.





















400 Nathan Nunn

La Porta, Rafael, Lopez-de-Silanes, Florencio, Shleifer, Andrei, Vishny, Robert, 1998. Law and finance.Journal of Political Economy 106, 1113–1155.

Liebowitz, S.J., Margolis, Stephen E., 1990. The fable of the keys. Journal of Law and Economics 33 (1),1–25.

Mamdani, Mahmood, 2001. When victims become killers: colonialism, nativism, and genocide in Rwanda.Princeton University Press, Princeton, N.J.

Mann,Charles C., 2011. 1493:Uncovering the NewWorld Columbus Created. Alfred A. Knopf,NewYork.Mehlum, Halvor, Moene, Karl,Torvik, Ragnar, 2003. Predator or prey? parasitic enterprises in economic

development. European Economic Review 47, 275–294.Michalopoulos, Stelios, 2012.The origins of ethnolinguistic diversity. American Economic Review 102 (4),

1508–1539.Michalopoulos,Stelios,Pappaioannou,Elias,2011.The Long-Run Effects of the Scramble forAfrica,NBER

Working Paper 17620.Michalopoulos,Stelios,Pappaioannou,Elias,2013. Pre-colonial ethnic institutions and contemporaryAfrican

development. Econometrica 81 (1), 113–152.Miguel, Edward, Roland, Gérard, 2011. The long run impact of bombing Vietnam. Journal of Develop-

ment Economics 96 (1), 1–15.Mokyr, Joel, 2008. The institutional origins of the industrial revolution. In: Helpman, Elhanan (Ed.), Insti-

tutions and Economic Performance, Harvard University Press, Cambridge, MA, 64–119.Murdock, George Peter, 1959. Africa: Its Peoples and Their Cultural History. McGraw-Hill Book

Company, NewYork.Murphy, Kevin M., Shleifer, Andrei,Vishny, Robert W., 1993. Why is rent-seeking so costly to growth.

American Economic Review Papers and Proceedings 83 (2), 409–414.Naritomi, Joana, Soares, Rodrigo R., Assuncao, Juliano J., 2012. Institutional development and colonial

heritage within Brazil. Journal of Economic History 72 (2), 393–422.Nisbett, Richard E., Cohen, Dov, 1996. Culture of Honor: The Psychology of Violence in the

South. Westview Press, Boulder.Nunn, Nathan, 2007. Historical legacies: a model linking Africa’s past to its current underdevelopment.

Journal of Development Economics 83 (1), 157–175.Nunn, Nathan, 2008a. The long-term effects of Africa’s slave trades. Quarterly Journal of Economics 123

(1), 139–176.Nunn,Nathan,2008b. Slavery, inequality,and economic development in theAmericas:an examination of the

Engerman-Sokoloff hypothesis. In: Helpman, Elhanan (Ed.), Institutions and Economic Performance.Harvard University Press, Cambridge, MA, 148–180.

Nunn, Nathan, 2009.The importance of history for economic development. Annual Review of Economics1 (1), 65–92.

Nunn, Nathan, 2012. Culture and the historical process. Economic History of Developing Regions 27,108–126.

Nunn, Nathan, Puga, Diego, February 2012. Ruggedness: the blessing of bad geography in Africa. Reviewof Economics and Statistics 94 (1), 20–36.

Nunn, Nathan, Qian, Nancy, May 2010. The columbian exchange: a history of disease, food, and ideas.Journal of Economic Perspectives 24 (2), 163–188.

Nunn,Nathan,Qian,Nancy,2011.The Potato’s contribution to population and urbanization:evidence froma historical experiment. Quarterly Journal of Economics 126 (2), 593–650.

Nunn, Nathan,Trefler, Daniel, forthcoming. Domestic institutions as a source of comparative advantage.In: Gopinath, Gita, Helpman, Elhanan, Rogoff, Kenneth (Eds.), Handbook of International Economics,vol. 4, North-Holland, NewYork.

Nunn, Nathan,Wantchekon, Leonard, 2011. The slave trade and the origins of mistrust in Africa. AmericanEconomic Review 101 (7), 3221–3252.

Nunn, Nathan, forthcoming. Gender and missionary influence in colonial Africa. In: Akyeampong,Emmanuel, Bates, Robert, Nunn, Nathan, Robinson, James A. (Eds.), Africa’s Development in His-torical Perspective.
























Olsson, Ola, 2004. Unbundling Ex-Colonies: A Comment on Acemoglu, Johnson, and Robinson, 2001.Goteborg University, Mimeo.

Olsson, Ola, Hibbs Jr., Douglas A., 2005. Biogeography and Long-Run Economic Development. EuropeanEconomic Review 49, 909–938.

Olsson, Ola, Paik, Christopher, 2012. A Western Reversal Since the Neolithic? The Long-Run Impact ofEarly Agriculture. University of Gothenburg, Mimeo.

Osafo-Kwaako, Philip, 2012. Legacy of State Planning: Evidence from Villagization in Tanzania. HarvardUniversity, Mimeo.

Peisakhin, Leonid, 2010. Living Historical Legacies: The “Why” and “How” of Institutional Persistence.Yale University, Mimeo.

Pierce, Lamar, Snyder, Jason A., 2012. Trust and Finance: Evidence from the African Slave Trade.University of California Los Angeles, Mimeo.

Puga,Diego,Trefler,Daniel,2012. InternationalTrade and Institutional Change:MedievalVenice’s Responseto Globalization. University of Toronto, Mimeo.

Putnam, Robert, Leonardi, Robert, Nanetti, Raffaella, 1993. Making Democracy Work. Simon & Schuster,NewYork.

Putterman,Louis,2008.Agriculture,Diffusion an Development:Ripple Effects of the Neolithic Revolution,Economica 75, 729–748.

Putterman, Louis, Weil, David N., 2010. Post-1500 population flows and the long-run determinants ofeconomic growth and inequality. Quarterly Journal of Economics 125 (4), 1627–1682.

Redding, Stephen J., Sturm, Daniel,Wolf, Nikolaus, 2011. History and industrial location: evidence fromGerman airports. Review of Economics and Statistics 93 (3), 814–831.

Reid,Richard, 2013.The fragile revolution: rethinking ware and development in Africa’s violent nineteenthcentury. In: Akyeampong, Emmanuel, Bates, Robert H., Nunn, Nathan, Robinson, James A. (Eds.),Africa’s Development in Historical Perspective,Cambridge University Press,Cambridge,p. forthcoming.

Rodrik,Dani,Subramanian,Arvind,Trebbi,Francesco,2004. Institutions rule:the primacy of institutions overgeography and integration in economic development. Journal of Economic Growth 9 (2), 131–165.

Rubin, Jared, 2011. Printing and Protestants: Reforming the Economics of the Reformation. CaliforniaState University, Fullerton, Mimeo.

Sokoloff, Kenneth L., Zolt, Eric M., 2007. Inequality and the evolution of institutions of taxation: evidencefrom the economic history of the Americas. In: Edwards, Sebastian, Esquivel, Gerardo, Márquez, Gra-ciela (Eds.),The Decline of Latin American Economies: Growth, Institutions, and Crises, University ofChicago Press, Chicago, 83–136.

Spolaore, Enrico,Wacziarg, Romain, 2009. The diffusion of development. Quarterly Journal of Economics124 (2), 469–529.

Spolaore, Enrico,Wacziarg, Romain, 2010. War and Relatedness. Tufts University, Mimeo.Spolaore, Enrico, Wacziarg, Romain, forthcoming. How deep are the roots of economic development?

Journal of Economic Literature.Tabellini, Guido, 2008. The scope of cooperation: values and incentives. Quarterly Journal of Economics

123 (3), 905–950.Tabellini, Guido, 2010. Culture and institutions: economic development in the regions of Europe. Journal

of the European Economic Association 8 (4), 677–716.Tai, En-Sai, 1918. Treaty Ports in China. Columbia University Press, NewYork.Tilly, Charles, 1990. Coercion, Capital and European States, A.D. 990–1990. Blackwell Publishers,

Cambridge.Vansina, Jan, 2004. The Antecedents of Modern Rwanda:The Nyiginya Kingdom. The University of Wis-

consin Press, Wisconsin.Verdier,Thierry, Bisin,Alberto, 2000. Beyond the melting pot: cultural transmission. marriage and the evo-

lution of ethnic and religious traits. Quarterly Journal of Economics 115, 955–988.Verdier, Thierry, Bisin, Alberto, 2001. The economics of cultural transmission and the dynamics of pre-

ferences. Journal of Economic Theory 97, 298–319.Voigtlaender, Nico,Voth, Hans-Joachim, 2012. Perpetuated persecution: the medieval origins of anti-semitic

violence in Nazi Germany. Quarterly Journal of Economics 127 (3), 1339–1392.Waldinger, Maria, 2012. Missionaries and Development in Mexico. London School of Economics, Mimeo.


















402 Nathan Nunn

Wang, Ke-Wen, 1998. Modern China: An Encyclopedia of History, Culture, and Nationalism. GarlandPublisher, NewYork.

Weber, Max, 1930. The Protestant Ethic and the Spirit of Capitalism. Routledge, London.Whatley, Warren, forthcoming. The trans-atlantic slave trade and the evolution of political authority in

West Africa. In:Akyeampong, Emmanuel, Bates, Robert H., Nunn, Nathan, Robinson, James A. (Eds.),Africa’s Development in Historical Perspective, Cambridge University Press, NewYork.

Wittfogel, Karl A., 1957. Oriental Despotism: A Comparative Study of Total Power. Yale UniversityPress, New Haven.

Woodberry, Robert D., 2004. The Shadow of Empire: Christian Missions, Colonial Policy, and Democracyin PostColonial Societies. PhD Dissertation in Sociology, University of North Carolonia at ChapelHill.

Woodberry, Robert D., 2012. The missionary roots of liberal democracy. American Political ScienceReview 106 (2), 244–174.

Zerbe, Richard O., Anderson, C. Leigh, 2001. Culture and fairness in the development of institutions inthe California gold fields. Journal of Economic History 61 (1), 114–143.






historicaldevelopment - harvard university · evidence,for how one important historical...

Documents